Climate change: Termites, fungi play more important role in decomposition than temperature

Climate change: Termites, fungi play more important role in decomposition than temperature

Climate change models could have a thing or two to learn from termites and fungi, according to a new study released this week.

For a long time scientists have believed that temperature is the dominant factor in determining the rate of wood decomposition worldwide. Decomposition matters because the speed at which woody material are broken down strongly influences the retention of carbon in forest ecosystems and can help to offset the loss of carbon to the atmosphere from other sources. That makes the decomposition rate a key factor in detecting potential changes to the climate.

But scientists from Yale, the University of Central Florida and SUNY Buffalo State found that fungi and termites, which help break down wood, may play a more significant role in the rate of decomposition than temperature alone.

The group’s findings appear in this week’s edition of the journal Nature Climate Change.

“The big surprise of this work was the realization that the impact of organisms surpassed climate as a control of decomposition across spatial scales,” said Joshua King, a biologist at UCF and co-author of the paper. “Understanding the ecology and biology of fungi and termites is a key to understanding how the rate of decomposition will vary from place to place.”

So how did scientists originally come up with temperature as the main factor in decomposition? It has to do with data and math. Scientists most often construct a model based on the average decomposition rates of sites that are in close proximity to each other. In this case, it appears that each local number matter because they reflect the activity of fungi and termites. The team suggests that scientists need to embrace the variability found across data collected from many different sites instead of averaging it all together to create better models with more accurate predictions.

The team reached this conclusion after running a 13-month experiment. They distributed 160 blocks of pine tree wood across five sub-regions of temperate forest in the eastern U.S. — from Connecticut to northern Florida — and then monitored the decay that occurred.

They selected similar forest types, hardwood deciduous forests, to focus on major differences in climate across the regional gradient. (The average annual temperature in southern New England is about 11 degrees Celsius cooler than Florida.) Within each of the five sub-regions they placed the wood blocks in different types of terrain to evaluate the effects of local versus regional factors as controls on decomposition.

“Most people would try to make sure everything was as standard as possible,” said Mark A. Bradford, an assistant professor of terrestrial ecosystem ecology at the Yale School of Forestry & Environmental Studies (F&ES) and lead author of the study. “We said, ‘Well, let’s generate as much variation as possible.’ So we put some blocks on south-facing slopes, where they would be warmer in the summer, and others on north-facing slopes where it’s colder. We put some on top of ridges and others next to streams where it was wetter.”

After 13 months, they measured how much wood had been lost, whether to the consumption of fungi growing on the wood or to termites consuming the wood.

According to their analysis, local-scale factors explained about three quarters of the variation in wood decomposition, while climate explained only about one quarter, contrary to the expectation that climate should be the predominant control.

“We’re reaching the wrong conclusion about the major controls on decomposition because of the way we’ve traditionally collected and looked at our data,” Bradford said. “That in turn will weaken the effectiveness of climate prediction.”

The team’s recommendation: collect more data at local sites and improve our understanding of how local conditions affect the organisms that drive decomposition, because they could significantly improve the effectiveness of climate change projections.

Story Source:

Materials provided by University of Central Florida. Note: Content may be edited for style and length.

Journal Reference:

  1. Mark A. Bradford, Robert J. Warren II, Petr Baldrian, Thomas W. Crowther, Daniel S. Maynard, Emily E. Oldfield, William R. Wieder, Stephen A. Wood, Joshua R. King. Climate fails to predict wood decomposition at regional scales. Nature Climate Change, 2014; DOI: 10.1038/nclimate2251

Termites help protect rainforests from climate change

In the first-ever experiment of its kind, by removing the majority of termites from a patch a forest in Borneo scientists were able to measure the impact that the insects have on the ecosystem.

But when a severe drought hit the study site at the beginning of the experiment, they discovered something unexpected. The research is published in the journal Science.В

The researchers, from the Museum and the University of Liverpool, initially set out just to measure the effect that termites had on rainforests in general.

Dr Paul Eggleton, Merit Researcher at the Museum and senior co-author of the study, says, ‘We’ve always known that in tropical rainforests ants and termites are really important players, but they have been underestimated as ecological agents in these systems.

‘I think this is because of a northern temperate bias. Temperate forests in the northern hemisphere have been studied much more thoroughly than tropical areas in the southern hemisphere. Because of this, everybody had assumed in the past that microbes are responsible for the majority of decomposition in tropical forests, because they are in the northern temperate regions.’

To demonstrate the insects’ vital role in tropical ecosystems, the researchers carried out the first-ever experimental removal of termites from a small patch of rainforest, using new methods to lower the number of termites in the process.

Termite chocolate

To pluck out the termites from the 2,500-metre-square plots, the researchers used over 4,000 toilet rolls.

‘Fortunately for us, termites love the cellulose parts of loo rolls,’ says Paul, referring to the white paper that surrounds the core. ‘It’s like giving them chocolate.

‘We were able to poison a whole host of loo rolls and put them out onto our plots, managing to reduce the termite activity by about three quarters.’

The poison has been rigorously tested to be certain that only termites are affected by it, resulting in the impressive three-quarter drop in activity.

This then allowed the researchers to measure the decomposition rate of wood and leaf litter, the distribution of nutrients in the soil, whether the termites had any effect on seedling survival, the moisture levels in the soil and also the depth of the leaf litter between plots with and without termites.

Termites play a vital role in maintaining the rainforests of Borneo В© Ben Houdijk/Shutterstock

‘For all of those things, we saw a very big difference between the plots where we had supressed termites and those where we hadn’t,’ says Paul. Leaf litter, for example, was 22% lower when termites were present, while seedling survival increased by over 50%.

While the huge benefit of termites to the forest was hardly unexpected, what happened during the experiment really took the team by surprise.

Extreme weather bites

It just so happened that as the experiment was ongoing, the Pacific Ocean experienced the largest El NiГ±o on record. This is the massive weather event that occurs in the Pacific every few years.

Normally, winds across the Pacific blow from South America towards Asia. This means that the warmer water builds up in the western side of the ocean while colder water is uplifted along the eastern edge. It is this temperature difference that drives the weather system.

During El NiГ±o, however, these winds are weakened, meaning the warm water that is usually found off the coasts of Asia and Oceania shifts further east, warming the waters around South America. This unbalances the weather systems and leads to drought across Indonesia and Borneo.

In 2015, while the termite experiments were in full swing the largest ever El NiГ±o hit Borneo plunging the rainforest into a severe drought.

‘At first, we thought the drought would ruin the experiment,’ explains Paul. ‘But much to our surprise we found that the termites did better in the drought conditions than in the wet conditions.

‘That was not what we were expecting at all.’

The termites help to protect the rainforests against drought by keeping soil moisture levels up В© Ihsan Saladin/Shutterstock

It turned out that while the researchers predicted that the dry weather would hinder all decomposition in the forest, the opposite was actually true. The termites massively ramped up their activity. Why this was the case is still not fully understood, but Paul thinks it is likely to do with how water-logged the soil is.

He explains, ‘They move around in tunnels in the soil, and so when it is very wet they can’t move about much, as many of their tunnels will be full of water. But when it is nice and dry they can make their tunnels more easily, resulting in much higher activity levels.’

Buffering climate change

The result is of this activity is not only an increase in the decomposition of leaf litter and wood, but also an increase in the overall level of moisture in the soil and the rate of nutrient cycling within the environment.

This has a knock-on effect for the plants growing in the rainforest, increasing the survival of seedlings by half during harsh droughts, and providing a mosaic of nutrients that in turn bumps up the overall biodiversity of the forest.

‘They mitigated the effect of drought,’ says Paul. ‘Of course, we’re expecting more droughts in the future, so systems that keep their natural level of termites should be provided with some defence.’

As more and more tropical rainforest is being cut down as the insatiable demand for resources grows, however, the outcome is not favourable for the termites.

According to Paul, even the slightest disturbance to a forest can have a serious impact on the number and diversity of termites that live there. It is now likely that this will affect the resilience of these forests to extreme weather events, which are only predicted to get more frequent and harsher in the future.

Are termites a legitimate contributor to climate change?

Are termites a legitimate contributor to climate change?

My father-in-law isn’t exactly a climate change-denier, but anytime my wife or I naively brings up global warming or strategies to minimize it, he instead notes that termites are a serious problem and asks why isn’t the government regulating these little pests more than it is other variables.

I do not have a science background whatsoever, so I often get stuck trying to regurgitate basic principles in climate change– but would love some insight or a point-in-the-right-direction as to whether or not his claim is real or if it’s just a strawman to delegitimize our insistence that global warming is a problem.

So there you have it: Are termites a legitimate contributor to climate change?

Thanks in advance, Internet friends.

Is he saying that termites are major contributors to climate change? If so, how does he say they contribute? Or is he merely just saying termites are a problem and if the government is reducing one they should reduce others too?

Thanks; I’ll have to talk to him more to get the full explanation.

From what I remember, though, his argument tries to discredit current or proposed environmental regulations by throwing up termites as another serious threat: “well termites actually are one of the biggest contributors to climate change, so why don’t we care about them as much as other variables that the media spotlights?”

From memories of my general education, termites produce large amounts of methane. Methane is a potent greenhouse gas, but whether termite mounds have any significant effect on local climate is not something that I have seen proved.

That’s where I get stuck, too. And thus how i ended up here!

From a bit of amateur journal scanning, it seems that termites do produce a disproportionate amount of methane relative to their size, but I was hoping to learn more about if their production is offset, whereas the release of fossil fuels is not.

Termites are indeed a significant source of the overall methane and CO2 budget. But termites have been around for billions of years. The issue with climate change is not the flows themselves, but that we’re changing the flows of greenhouse gasses in and out of the atmosphere.

Imagine the atmosphere as a simple sink-and-source model; essentially a bathtub with a spigot going in, and a drain going out. If the flow out is dependent on the depth of the water, if the spigot is on at a constant flow you will reach an equilibrium state where the flow out is equal to the flow in. If, however, we come and turn the flow in up, the water level will rise.

So in this scenario, we’ve been happily at one water level for a while, with inflows (from termites, and microbes, and decomposing vegetation, and volcanoes) in rough equilibrium with outflows (ocean uptake, plant growth, etc). When the water gets turned up by humans it will change the equilibrium state; the fact that the termites (and everything else) are still there doesn’t make them the source of the problem. We are the thing that is changing. Termites are just doing their little termite thing.

Global Impact of Termites on the Carbon Cycle and Atmospheric Trace Gases

  • Atsuko Sugimoto
  • David E. Bignell
  • Jannette A. MacDonald

Termites have high biomass in many tropical ecosystems and emit the greenhouse gases CO2 and CH4. They are also recognized as ecosystem engineers, mediating decomposition and other aspects of soil function. Therefore, termites may be significant contributors to biogeochemical cycles, notably those of carbon and methane. We review methods of assessing carbon fluxes through termite populations and argue that direct measurements of net CO2 and CH4 emissions from termites in natural settings (in their nests or in the soil) are the best data for scaling-up calculations, if accompanied by accurate estimates of biomass and assemblage feeding-group composition. Actual determinations of gas fluxes from termites, and the attendant computation of regional and global budgets made over the past two decades are reviewed. For CO2, it is concluded that termites contribute up to 2% of the natural efflux from terrestrial sources, a large contribution for a single animal taxon, but small in the global context. For CH4, we note that calculations are still hampered by uncertainties over termite biomass distribution and a general failure to consider local and landscape-level oxidation by methylotrophic microorganisms as a factor mitigating net fluxes. Nevertheless the balance of evidence, including new data on local oxidation, suggests that annual contributions by termites are almost certainly less than 20 Tg, and probably less than 10 Tg (ca. 4% and 2% of global totals from all sources, respectively). Climate changes and land use intensification may cause minor modifications of the overall distribution of termites, but a more serious impact on soil stability and function could result from changes in the balance of feeding groups. The response of termites to changes in the quality and quantity of plant litters is uncertain, but direct effects from elevated atmospheric CO2 are unlikely. Global changes will broadly favour wood- and litter-feeding termites over soil-feeders, but with regional differences and complications arising from patterns of landscape fragmentation and historical factors.

Unable to display preview. Download preview PDF.

Invasive termites in a changing climate: A global perspective

Grzegorz Buczkowski

1 Department of Entomology, Purdue University, West Lafayette, IN, USA

2 Purdue Climate Change Research Center, Purdue University, West Lafayette, IN, USA

Cleo Bertelsmeier

3 Department of Ecology and Evolution, Biophore, UNIL‐Sorge, University of Lausanne, Lausanne, Switzerland

Associated Data

Termites are ubiquitous insects in tropical, subtropical, and warm temperate regions and play an important role in ecosystems. Several termite species are also significant economic pests, mainly in urban areas where they attack human‐made structures, but also in natural forest habitats. Worldw >RCP s), RCP 4.5 and RCP 8.5, and two projection years (2050 and 2070). Our results show that all but one termite species are expected to significantly increase in their global distribution, irrespective of the climatic scenario and year. The range shifts by species (shift vectors) revealed a complex pattern of distributional changes across latitudes rather than simple poleward expansion. Mapping of potential invasion hotspots in 2050 under the RCP 4.5 scenario revealed that the most suitable areas are located in the tropics. Substantial parts of all continents had suitable environmental conditions for more than four species simultaneously. Mapping of changes in the number of species revealed that areas that lose many species (e.g., parts of South America) are those that were previously very species‐rich, contrary to regions such as Europe that were overall not among the most important invasion hotspots, but that showed a great increase in the number of potential invaders. The substantial economic and ecological damage caused by invasive termites is likely to increase in response to climate change, increased urbanization, and accelerating economic globalization, acting singly or interactively.

1. Introduction

The spread of exotic species, climate change, and urbanization are among the most serious global environmental threats. Each factor is independently capable of effecting significant changes in biological communities, and all three have been the subject of extensive research in the context of conservation and the control of pests (e.g., Dukes & Mooney, 1999; Hudson et al., 2014; Sala et al., 2000; Walther et al., 2009). More recently, studies investigating the connectedness of these factors and their potential cumulative interactions have become more common (e.g., Brook, Sodhi, & Bradshaw, 2008; Buczkowski & Richmond, 2012; Gallardo & Aldridge, 2013; Mooney & Hobbs, 2000; Stachowicz, Terwin, Whitlatch, & Osman, 2002). Although such studies are still relatively rare, the synergy between these issues is becoming increasingly evident. For example, changing climatic conditions are expected to alter global commerce routes in the future and likely increase the introduction of exotic species into new geographic regions (Bradley, Blumenthal, Wilcove, & Ziska, 2010; Hellmann, Byers, Bierwager, & Dukes, 2008).

While the degradation of ecosystem services and biodiversity by invasive species is already a major challenge, climate change is likely to increase it. There is a general consensus that the future distribution of invasive species will likely expand with climate change (Bellard et al., 2013; Dukes & Mooney, 1999; Mooney & Hobbs, 2000; Ziska & Dukes, 2014). Previous studies have shown that changes in broad climatic conditions may influence the probability of species invasions and that such effects are likely to be diverse and context‐dependent (Bradley et al., 2010; Rahel & Olden, 2008; Walther et al., 2009). In comparison with native species, invasive species are more likely to adapt to the new climatic conditions because they are usually abundant, tolerate a broad range of climatic conditions, cover wide geographic ranges, and have highly competitive biological traits (Hellmann et al., 2008). Humans inadvertently transport a wide range of species around the globe, and although many of these inoculations presumably fail because of inhospitable climate in the recipient region (Williamson & Fitter, 1996), global warming may relax this constraint. This may especially be true for insects, which are dependent on external sources of body heat (ectotherms), and whose spread has formerly been restricted by climatic barriers.

Among insects, the highly advanced eusocial societies of ants (Hymenoptera: Formicidae) and termites (Dictyoptera: Termitidae) have been especially problematic as invaders in natural, urban, and agricultural ecosystems (reviewed in Holway, Lach, Suarez, Tsutsui, & Case, 2002; Evans, Forschler, & Grace, 2013). Previous studies have modeled the potential spread of invasive ants under climate change and demonstrated that a large amount of global landmass is climatically suitable to ant invasions (Bertelsmeier, Guenard, & Courchamp, 2013; Bertelsmeier, Luque, Hoffmann, & Courchamp, 2013, 2015). However, climate change and ant invasions were not predicted to act synergistically and the impacts on invasive ants were expected to either increase or decrease depending on the taxon (Bertelsmeier, Blight, & Courchamp, 2016). Furthermore, the ant invasion hotspots were predicted to occur mainly within biodiversity hotspots (Bertelsmeier et al., 2015), which is especially problematic for biodiversity conservation.

Despite the economic and ecological importance of invasive termites, no study has modeled their potential global distribution under climate change. Termites are cryptic social insects that play an important role in the carbon cycle and act as important ecosystem engineers in most of the world’s tropical ecosystems. They contribute to the carbon cycle by feeding on a wide range of living, dead, and decaying plant matter (Bignell & Eggleton, 2000; Traniello & Leuthold, 2000), by comminution of wood and other plant residues, and by modifying soil physical properties such as texture, water infiltration rates, and nutrient contents at various spatial scales (e.g., Dangerfield, McCarthy, & Ellery, 1998). Termites are widely distributed throughout the tropical and subtropical regions of the world (Eggleton, 2000), with the highest diversity found in tropical forests where they comprise the greater part of insect biomass (Bignell & Eggleton, 2000). Despite the ecological benefits of termites, they are also significant pests causing damage to human‐built structures (Su & Scheffrahn, 1998) and tropical agriculture (Rouland‐Lefèvre, 2011). In contrast to the well‐known ecological effects of other invasive social insects such as ants (Holway et al., 2002), the ecological consequences of termite invasions remain poorly understood and most research has focused on economic consequences in urban areas.

Worldwide, the number of recognized invasive termite species has increased from 17 in 1969 to 28 today and invasive termites are increasing in both number and geographic area (Evans et al., 2013). A single recent study attempted to predict the potential habitat of Coptotermes formosanus and Coptotermes gestroi in Florida using occurrence data and climate modeling (Tonini, Divino, Lasinio, Hochmair, & Scheffrahn, 2014), but a global assessment of a wider range of invasive termite species is lacking. The goal of the current project was to provide a global risk assessment for invasive termites under scenarios of climate change using 13 of the most aggressive pest species. We model suitable area globally for these 13 invasive termite species, both currently and with predicted climate change (in 2050 and 2070). Such research is crucial for identifying areas with the highest risk of invasions and for implementing proactive management responses in the case of invasions.

2.1. Species distribution data

Worldwide, approximately 28 termite species are considered invasive (Evans et al., 2013) and we selected 13 to include in the global projection of termite invasion risks. These species were selected based on a number of factors. First and foremost, we selected species that are the most economically and ecologically important. For example, the Formosan subterranean termite (C. formosanus) and the Asian subterranean termite (C. gestroi) are the two most destructive termite pests in the world and are responsible for most of the $40 billion annual economic impact from termite damage (Evans et al., 2013). Coptotermes formosanus is on the list of the “100 of the world’s worst invasive species” (Lowe, Browne, Boudjelas, & De Poorter, 2000). The eastern subterranean termite (Reticulitermes flavipes) is native to the eastern United States, but has spread to various parts of the world including Europe, South America, and several oceanic islands (Dronnet, Chapuisat, Vargo, Lohou, & Bagneres, 2005). It is the most common and the most economically important termite in the United States and is responsible for approximately $2 billion in damage annually (Su & Scheffrahn, 1990). Similarly, the highly destructive West Indian drywood termite (Cryptotermes brevis), native to coastal deserts in Peru and Chile, has invaded all continents and numerous oceanic islands is more frequently introduced into new locations than any other termite in the world (Evans et al., 2013). Second, we selected species for which occurrence data in both native and introduced ranges have been adequately described. Termite taxonomy and species identification have been problematic for a long time, and only recently molecular diagnostic tools have been used to answer questions about the sources and sinks of invasive termites (Evans et al., 2013). For example, Reticulitermes santonensis was considered native to France, in part because it is found in forests there. However, mitochondrial DNA sequence data have shown that R. santonensis is an invasive population of R. flavipes (Austin et al., 2005), a native of southern United States introduced into France before 1840 (Bagneres et al., 1990).

Based on the above criteria we selected 13 species: C. formosanus, C. gestroi, C. brevis, Cryptotermes cynocephalus, Cryptotermes dudleyi, Cryptotermes domesticus, Cryptotermes havilandi, Incisitermes immigrans, Incisitermes minor, Mastotermes darwiniensis, Nasutitermes corniger, R. flavipes, and Reticulitermes grassei. The distribution records for the 13 species were obtained from various sources including the primary literature (reviewed in Evans, 2010; Jones & Eggleton, 2011; Evans et al., 2013), the IUCN database for invasive species (IUCN SSC Invasive Species Specialist Group 2012; Jones & Eggleton, 2011), and CABI’s Invasive Species Compendium (CABI 2016).

Because the models should include the full range of environmental conditions under which the species can thrive, we included occurrence points from both the native and the invaded range (following Beaumont et al., 2009; Broennimann et al., 2007; Liu, Guo, Ke, Wang, & Li, 2011). It has been shown that models calibrated on native range data alone often misrepresent the potential invasive distribution and that these errors propagate when estimating climate change impacts (Beaumont et al., 2009; Broennimann et al., 2007).

We used on average 42 occurrence points to model the species’ distribution (46 points for C. formosanus, 61 for C. gestroi, 110 for C. brevis, 20 for C. cynocephalus, 44 for C. dudleyi, 42 for C. domesticus, 38 for C. havilandi, 21 for I. immigrans, 36 for I. minor, 20 for M. darwiniensis, 40 for N. corniger, 40 for R. flavipes, and 20 for R. grassei). In order to make robust range predictions, it is not necessary to include every single location where the species is present, but a representative cover of all climatic conditions under which the species is known to live should be included. Our occurrence records come from all continents (except Antarctica where termites do not occur) and include tropical and temperate locations, over a wide range of latitudes. Nonetheless, we excluded species with less than 20 occurrence points (see Franklin, 2009). As all the chosen modeling methods also require absence data, we generated three sets of 1,000 randomly selected pseudo‐absences with equal weighting for presences and absences (Barbet‐Massin, Jiguet, Albert, & Thuiller, 2012).

2.2. Climatic predictors

To construct and project SDMs predicting the current potential distribution of the 13 termite species, we used bioclimatic variables from the Worldclim database, which represent averaged values over the period 1950–2000 (Hijmans, Cameron, Parra, Jones, & Jarvis, 2005). Previous studies on climatic niches of species and biological invasions have used these variables (Wolmarans, Robertson, & van Rensburg, 2010). Instead of simply using monthly data on temperature or rainfall, which may not have a particular significance to the organism, these variables represent derived metrics (Hijmans et al., 2005) that are known to influence species distributions (e.g., temperature of the warmest quarter) (Root, Price, & Hall, 2003). The bioclimatic variables represent annual trends (e.g., annual precipitation), limiting environmental factors (e.g., temperatures of the coldest month), and seasonality (e.g., annual range in temperature and precipitation) (Hijmans et al., 2005). The spatial resolution of the GIS layers was approximately 18.5 × 18.5 km (10 arcmin).

Termite ecophysiology is insufficiently well developed to identify individual limiting environmental factors for each species, although temperature and humidity are certainly important (Clarke, Thompson, & Sinclair, 2013). We selected three variables for each species using a three‐step procedure: (1) We tested the variable importance using the variable selection procedure in the Biomod2 package and averaged relative variable importance across all available algorithms in this package, (2) we assessed pairwise correlations among all 19 bioclimatic variables, and (3) we selected the three most important uncorrelated variables (Pearson’s r 1 for variable selection per species and the relative contribution of the variables averaged across all models). We used GIS layers with climatic change data of future scenarios using the 5th IPCC assessment report (IPCC 2014). The WorldClim database provides projections that are downscaled to the same spatial resolution as the data for “current” conditions. Future climate scenarios are based on different geophysical hypotheses of how the Earth’s climate will react to the increase in the amount of greenhouse gases. Therefore, we used a range of three different geophysical global circulation models (GCMs), which simulate the climate in response to different socioeconomic storylines: the GISS‐ES‐R model; the HadGEM2‐ES model; and the MIROC‐ESM model (IPCC 2014). To account for different socioeconomic scenarios, we used two different Representative Concentration Pathways (RCPs), which represent a midrange (RCP 4.5: +1.1–2.6°C by the year 2100) and a more pessimistic scenario (RCP 8.5: +2.6–4.8°C by the year 2100).

Selected variables and their relative importance (average contribution to the models in %) per species and modeling algorithm

Variables Bioclim code Cbre Ccyn Cdom Cdud Cfor Cges Chav Iimm Imin Mdar Ncor Rfla Rgra
Annual mean temperature Bio1 27.20 17.97
Mean diurnal range Bio2 33.42 26.73 10.53 29.49
Isothermality Bio3 24.63 49.11
Temperature seasonality Bio4 38.51
Max temperature of warmest month Bio5 23.49 11.18
Min temperature of coldest month Bio6 60.27 55.67 50.15
Temperature annual range Bio7 68.39
Mean temperature of wettest quarter Bio8 26.62 61.87 39.46
Mean temperature of driest quarter Bio9
Mean temperature of warmest quarter Bio10 35.15 30.00 23.58
Mean temperature of coldest quarter Bio11 60.55 27.47 51.47 43.77
Annual precipitation Bio12 23.54
Precipitation of wettest month Bio13 16.24 49.80 13.64 10.39
Precipitation of driest month Bio14 6.03
Precipitation seasonality Bio15 13.50
Precipitation of wettest quarter Bio16 45.80 33.15 21.40
Precipitation of driest quarter Bio17 32.68
Precipitation of warmest quarter Bio18 37.65 31.48
Precipitation of coldest quarter Bio19 38.00

Species abbreviations in the top row are as follows (from left to right): Cryptotermes brevis, Cryptotermes cynocephalus, Cryptotermes domesticus, Cryptotermes dudleyi, Coptotermes formosanus, Coptotermes gestroi, Cryptotermes havilandi, Incisitermes immigrans, Incisitermes minor, Mastotermes darwiniensis, Nasutitermes corniger, Reticulitermes flavipes, Reticulitermes grassei.

2.3. Species distribution modeling

We used 10 statistical and machine learning methods to model the climatic niche of the 13 termite species under current and future (2050 and 2070) climatic conditions. The models were calibrated and projected using the BIOMOD2 package v.3.3.7 (Thuiller, Lafourcade, Engler, & Araújo, 2009) and included (1) generalized linear models (GLM), (2) generalized additive models (GAM), (3) generalized boosted models (GBM), (4) classification tree analysis (CTA), (5) flexible discriminant analysis (FDA), (6) multivariate adaptive regression splines (MARS), (7) random forests (RF), maximum entropy (Maxent), (9) surface range envelopes (SRE), and (10) artificial neural networks (ANN).

To validate the models, we performed 10‐fold cross‐validation. At each run, 70% of the occurrence points are selected at random and then used to train the models and the remaining 30% of occurrence points are kept for model evaluation (Guisan & Thuiller, 2005). To test predictive performance, we used with two metrics: the area under the receiver operating characteristic curve (AUC) (Fielding & Bell, 1997) and the true skill statistic (TSS) (Allouche, Tsoar, & Kadmon, 2006).

A clear limitation of species distribution modeling is that any particular prediction is contingent on the model input data. Yet, multiple sources of uncertainty create a variety of potential outputs (Buisson, Thuiller, Casajus, Lek, & Grenouillet, 2010). Here, we base our predictions on several modeling methods, global circulation models, and socioeconomic storylines. One way to deal with this “noise” and to try to filter out a signal from these multiple forecasts is to conduct consensus forecasts (Araújo & New, 2007), which superpose individual forecasts. Here, we combined models using the ten different modeling techniques with each of the three global climate models (GCM).

As individual models can vary in their predictive accuracy, their contribution to the final consensus forecasts was weighted according to their TSS. We used only the binary predictions and not the suitability indices of the individual model outputs to create the consensus prediction because continuous outputs of different modeling methods can be probabilities or indices with different mathematical meanings (Guo & Liu, 2010). However, adding individual presence–absence predictions spatially, and scaling the value to 1, produces a suitability index that can indeed be interpreted as the probability that the grid cell presents favorable environmental conditions for the species (Araújo & New, 2007).

We generated consensus models under current climatic conditions (over 10 modeling methods), and for future climatic conditions (over 10 modeling methods and three global circulation models). For future climatic conditions, this yielded a separate consensus projection per year (2050 and 2070) and socioeconomic pathway (RCP). We also calculated the standard error of the mean between climatic scenarios in order to show the extent of variation across forecasts (Barbet‐Massin, Rome, & Muller, 2013).

2.4. Assessing suitable area

To assess the total suitable area for each species and the changes in suitable area with climate change, we converted the consensus projections into binary (presence–absence) predictions) using the binary transformation function in Biomod2. We stacked the binary presence–absence predictions of the 13 species in order to create “invasion hotspot” maps. We then created invasion hotspot delta maps by subtracting the current hotspot map from the future hotspot map, showing pixels that are predicted to lose or gain potential invaders. We also mapped predicted range shifts for each of the 13 species showing gained, lost, and stable habitat under future climatic conditions. To assess whether the range margins have contracted or expanded, we calculated shift vectors of the range margins in all four cardinal directions (15% of the most extreme points in either direction) and we also calculated a shift vector for the center of gravity of the species distribution. Using the s.table() function in the ade4 package, we graphically compared the sizes of the different range shift vectors, to assess whether species shift preferentially in one particular direction and whether distributional changes are predominantly expected at the range margins.

Most models showed fair to very good performance (Table 2 ), and those with insufficient TSS scores were discarded. Following climate change, almost all species (12 of 13) showed an increase in potential range size under both socioeconomic development scenarios and for both projection years. In 2050, under the RCP 4.5 scenario, all species were predicted to increase: C. brevis (+7.5%), C. cynocephalus (+10.1%), C. domesticus (+20.3%), C. dudleyi (+3%), C. formosanus (+16%), C. gestroi (+4%), C. havilandi (+6%), I. minor (+2.7%), M. darwiniensis (+54.2%), N. corniger (+3.5%), R. flavipes (+16.7%), R. grassei (25%), with the exception of I. immigrans which was predicted to slightly decrease (−2.8%). Under the RCP 8.5 scenario and for the year 2070, the projections were of similar magnitude (Figure 1 ).

Change in potential range size (%) according to two socioeconomic storylines ( RCP 4.5 and RCP 8.5) in 2050 and 2070

No Comments

Leave a Reply

Your e-mail will not be published. All fields are required.