Local species richness may be declining after all

Recently two papers seemed to turn what we thought we knew about changes in biodiversity on their head. These papers by Vellend et al. and Dornelas et al. collated data from multiple sources and suggested that species richness at local scales is not currently declining. This was counter-intuitive because we all know that species are going extinct at unprecedented rates. However, it is possible that the introduction of non-native species and recovery of previously cultivated areas may offset extinctions leading to relatively little net change in local species richness.

This week a paper has been published that calls these findings into question. The paper by Andy Gonzalez and colleagues published in the journal Ecology, suggests that there are three major flaws with the analyses. These flaws mean that the answer to the question ‘Is local-scale species richness declining?’ currently remains unanswered and is unanswerable.

The papers of Vellend et al. and Dornelas et al. were meta-analyses of previously published papers. One issue with meta-analysis is that it is very prone to bias. Like any study if the samples (in this case ecological studies) are not representative of the population (in this case locations around the globe) then any results will be flawed. To test the representativeness of the datasets used by Vellend and Dornelas Gonzalez et al. examined how well they represented biodiversity and threats to biodiversity. This analysis (see below) showed that the papers were not representative of biodiversity or the threats faced by biodiversity (though curiously, the analysis of Dornelas et al. showed an overrepresentation of areas highly impacted by human impacts).

Gonzalez_1.png
Figure 1 – Spatial bias of the Vellend et al. (2013) and Dornelas et al. (2014) data syntheses. For more information see the paper by Gonzalez et al. (2016).

The paper also suggests that using short time series can underestimate losses. By analysing the effect of study duration and changes in species richness (see below) Gonzalez et al. claim that increases in study duration were correlated with a decline in species richness. This supports previous theory which suggests that there is often a time lag between disturbance events and species extinctions – termed ‘extinction debt.’ However, I’d be intrigued to see the results of removing the studies with the longest duration from this analysis since the authors admit that the analysis is sensitive to their inclusion. I’ve seen recent similar work that suggests the same kind of relationship might be seen for studies monitoring individual animal populations.

Gonzalez_2
Figure 2 – The effect of study duration on apparent changes in species richness.

Thirdly, Gonzalez et al. assert that including studies in which ecosystems were recovering from disturbance (e.g. regrowth on former agricultural fields) without taking into account historical losses that occurred during or after the disturbance biases estimates of change. The paper by Vellend et al. in particular combined studies of the immediate response of biodiversity to disturbances such as fire and grazing along with studies of recovery from the very same disturbances. Gonzalez et al. show that once studies of systems that were recovering are removed from Vellend et al’s analysis there is a negative trend in species richness changes.

The biases prevalent in the Vellend and Dornelas papers lead to Gonzalez et al. to suggest that the papers cannot conclude what the net changes in local species richness are at a global scale. However, they note that the results of Dornelas and Vellend are in sharp contrast to other syntheses of biodiversity changes which used reference undisturbed such as those by Newbold et al. and Murphy and Romanuk which reported average losses of species richness of 14 and 18% respectively.

In their conclusion Gonzalez et al. suggest that though meta-analysis is a powerful tool, it needs to be used with great care. Or to put it another way, with great power comes great responsibility. As someone who regularly uses meta-analysis to form generalisations about how nature works I completely agree with this statement. Traditionally scientists have used funnel plots (graphs with study sample size on the y-axis and effect size on the x-axis) to identify biases in their analyses. I’ve always been skeptical of this approach, especially in ecology where there is always a large amount of variation between sites. In the future syntheses would do well to follow the advice of Gonzalez et al. and really interrogate the data they are using to find any taxonomic, geographic, climatic or any other biases that might limit their ability to generalise. I know it’s something I’ll be taking more seriously in the future.

Gonzalez et al. also point out that most ecological research is carried out in Europe and North America. If we want to monitor biodivesity we need to increase efforts in biodiverse tropical regions, as well as boreal forests, tundra and deserts. We need to identify where these gaps need filling most and then relevant organisations need to prioritise efforts to carry out monitoring. I am positive that this can be achieved, but it will cost a lot money, needs to be highlighted as a priority and will ned a lot of political good will. Even with this effort some of the gaps in biodiverse regions, such as the Democratic Republic of Congo, will be extremely difficult to fill due to ongoing armed conflict

My take-home message from this paper is that we need to be more careful about how we do synthesis. However, I also think that species richness isn’t the only metric that we should focus on when talking about biodiversity change. Studies have shown that measures of the traits of species present in a community are generally more useful for predicting changes in ecosystem function than just using species richness. Species richness is the iconic measure of biodiversity, but it probably isn’t the best. Ecologists should view species richness in the same way as doctors view a thermometer – it’s a useful tool but you still need to be able to monitor blood pressure, take biopsies and listen to a patient’s lungs before you diagnose them*.

 


 

*Thanks to Falko Bushke whose analogy I stole from a comment he made on my blog post here.

 

Advertisements

Beta-diversity – What is it good for?

A while ago I wrote a post asking whether everyone’s favourite measure of biodiversity, species richness, was useful. In it, I concluded that it is probably one of the bluntest, least informative measures of ecological communities we have and that we should try to use alternative metrics when possible. Recently, I started wondering about what other measures of biodiversity might be informative, and what they can be used for. And then a neat review of beta-diversity by James Jacob Socolar ( correction courtesy of James Gilroy on Twitter – thanks James!) and colleagues came out in Trends in Ecology and Evolution, so today I’ll focus on that, borrowing from some of their thoughts and hopefully adding some of my own along the way. In the future, at some point, I’ll write something about temporal changes in ecological communities at individual sites.

So, firstly what do I mean by beta-diversity? Beta-diversity broadly reflects the differences in community composition between sites.  Gamma diversity (regional diversity) is a product of both beta- and alpha-diversity (diversity at a single site). And there are lots* of different ways of measuring beta-diversity. The simplest metric for beta-diversity is termed ‘true beta-diversity’ and was defined by Whittaker in 1960 as:

\beta=\frac\gamma\alpha

This metric is perhaps the easiest to interpret, but it also needs a reliable estimate of gamma diversity, so may be difficult to use in practice. Using this method allows the relationship between alpha and gamma diversity to be investigated. Other measures can be based on dissimilarity matrices, identifying pairwise differences between sites. These metrics can then be used to look at drivers of these differences, such as the geographic distance between individual sites and environmental differences. However, dissimilarity matrix methods don’t allow the relationship between alpha and gamma diversity to be investigated. The above explanation probably explains the ubiquity of species richness as a metric in ecology – we can all (more-or-less) agree on what it means.

Changes in beta-diversity when humans alter natural landscapes can be unpredictable. When human disturbances are patchy, such as in the case of selective logging, beta diversity has been shown to be stable or increase due to an influx of generalist species in forest gaps.

berry_et_al
Differences changes in tree community dissimilarity with increasing distance between sites in unlogged and logged forest. Note that logged forests show a more rapid rate of change, suggesting that logging results in more variable ecological communities. Figure modified from Berry et al 2008.

In contrast, when human land-use change results in the conversion of natural ecosystems to a relatively homogeneous system in which only a small subset of species can survive, beta-diversity tends to decrease. Examples of such drivers include agricultural conversion and urbanisation. However, even high intensity farming can result in an increase in beta-diversity particularly if species populations decrease leading to greater dissimilarity purely as a result of random processes.  In summary, the response of beta-diversity to anthropogenic change appears to be relatively idiosyncratic.

All of this is well and good, but what use is beta-diversity to practical conservation? At first inspection, this is not clear. The general perception of species richness is that more species = better**. Does higher beta-diversity = better? Well, no, not necessarily. Given that the aims of conservation vary from place to place, it is not surprising that how beta-diversity can be used also varies.

The most obvious use of beta-diversity is in spatial planning of protected areas. In landscapes which show a high spatial turnover of species, managers might favour the use numerous distinct reserves to capture this variation. However, in a landscape in which beta-diversity results from differents in species richness a single protected area might be favoured. Also, if a natural ecosystem is particularly distinct from other candidate sites it may be considered a priority for protection.

High beta-diversity can also result from dispersal limitation in a landscape. For example, secondary forests in fragmented landscapes plants with seeds dispersed by wind may colonise sites more readily than those dispersed by animals that may not cross non-forest areas. So in cases where beta-diversity amongst patches of a similar habitat in a fragmented landscape is high, this may point to the need for restoration to increase connectivity. Successful restoration may result in a decrease of beta-diversity as dispersal between patches increases. For example, Renata Pardini’s work has shown that the small mammal communities of more highly connected fragments of Atlantic forest are more similar to other patches than unconnected fragments. However, as far as I know, there is relatively little evidence empirical that restoration has similar effects.

In the paper I mentioned earlier, Jacob Socolar and colleagues suggest that beta-diversity may also be useful in informing the land-sharing vs land-sparing debate (which i have previously written about here, here an here). They argue that the use of beta-diversity as part of this debate may show that heterogenous landscapes that include agri-environment schemes, management of natural systems and high intensity agriculture are better at maintaining alpha- beta and gamma-diversity. Thus, the incorporation of metrics other than population sizes of species, the classic approach for such comparisons, may produce different conclusions to current studies, which largely suggest land-sparing as a favoured approach. As always with conservation, this depends on what you think we should try to protect. Should we focus on particular species? Or should we look attempt to conserve the processes that maintain coarse-scale diversity?

For me, the key point that the paper makes is that even though two recent high-profie studies recently suggested local-scale alpha-diversity is relatively constant***, global scale gamma-diversity is declining. This suggests that rare species are getting rarer and common species are increasing in abundance. If we can work out how and why beta-diversity responds to land-use changes we can better understand how to conserve gamma-diversity. However, before we do that we need to develop methods to upscale from alpha to gamma diversity and determine how different disturbances alter beta-diversity. Novel approaches offer the potential to solve this problem, but substantial testing is needed to determine how useful they are.


*Patricia Koleff identified 24 metrics for use with presence-absence data and my  old CEH office mate Louise Barwell tested 29 different beta-diversity metrics that incorporated abundance data. Give both of these papers a read, they’re well worth your time.

**I don’t agree with this perception, I’m just extrapolating based on things I have heard from a few people. Deeply unscientific, I know.

***I saw Andrew Gonzalez present some work on the problems of these two studies at the 2015 British Ecological Society annual meeting and hope to post something when the paper comes out. I can’t say much, but it was fascinating stuff.

 

New paper: Stand dieback and collapse in a temperate forest and its impact on forest structure and biodiversity

We recently published the first paper from my post-doc in Forest Ecology and Management, so I thought I’d share it here. It marks a bit of shift away from the tropical forests I have previously published about (see posts on that here and here), but it allowed me to continue my work on post-disturbance recovery.


Scientists and policymakers around the world are concerned about the potential effects of forest dieback. Drought and the spread of new pathogens and pests have resulted in increased tree mortality in both the USA and Canada, and these threats are likely to increase in Europe as well. The IPCC recently highlighted forest dieback as a potential major threat, but one about which we know relatively little.  Changes in forest biodiversity and ecosystem services are likely to be particularly severe in ecosystems that show poor resilience. Failure to withstand or recover from drought or pest attack may lead to ‘regime shifts’ resulting in a very different type of system, with many fewer trees.

Luckily for our group my boss, Adrian Newton, found out about a permanent transect that had been set up in the 1950s in a woodland in the New Forest that now appears to be suffering from dieback. The site had been surveyed 4 times between 1964 and 1999, and our team collected more data from the site in 2014. In our recent paper, published in Forest Ecology and Management we used this data to investigate dynamics of the woodland. In particular, we addressed the potential impacts of dieback on forest structure, the causes of these changes and their impact on biodiversity.

Basal area loss in Denny wood from 1964-2014
Basal area loss in Denny wood from 1964-2014

To cut a long story short, the forest lost about a third of its basal area (as you can see above) and over two-thirds of its juvenile trees over 50 years. over 90% of the loss of basal area was due to the death of large beech (Fagus sylvatica) and oak (Quercus rubor) trees.

Climate records from 1964-2014 showed that (a) mean temperature during April-September increased from 1960s to present day; and (b) there were numerous drought yearspost 1976.
Climate records from 1964-2014 showed that (a) mean temperature during April-September increased from 1960s to present day; and (b) there were numerous drought years post-1976 a year which was previously identified as a cause of current mortality.

The external factors causing these changes are not entirely clear, but there have been a number of significant droughts between 1964-2014 as well as increased temperatures (see figures above). In addition, the presence of a number of novel fungal pathogens has been noted in the forest, which may have interacted with drought to further weaken large trees. Recovery in the forest has been very limited, with almost no recruitment of saplings of the canopy dominants (beech & oak) in 50 years. This low recruitment is probably a result of the high density of ponies and deer in the woodland.

Relationship between percentage loss in subplot basal area and (a) percentage grass cover and (b) ground flora species richness.
Relationship between percentage loss in subplot basal area and (a) percentage grass cover and (b) ground flora species richness.

The result of the changes in forest structure is that areas with little tree cover have seen large increases in grass cover and increased ground flora species richness (see figure above). Both of these results indicate that there may be a tipping point at which changes in structure result in rapid increases in grass cover and species richness of ground flora.

Many of the papers on resilience talk about alternative stable states, in which transitions from one type of system to another are difficult to reverse. Though, from the outside, it may appear that our field site shows evidence of a shift to a relatively treeless stable state, we think that this is incorrect. The theory underlying multiple stable states suggests that disturbances causing the regime shift should be a ‘pulse’, when disturbance occurs over a relatively short period and then does not occur again, rather than a ‘press’ disturbance, where the disturbance is present over long periods of time.  However, these conditions are not met by our site where both pulse (i.e. drought) and ongoing press (i.e. overgrazing)  disturbances are present. We think that both of these processes are needed to cause the forest to lose tree cover.

Even if the transition we  have observed is not strictly a ‘regime shift’ it’s still important. Dieback is apparently widespread in the New Forest and is on-going, so the potential impacts could be very significant. As with other cases of dieback it’s difficult to identify appropriate management responses. However, in the case of the New Forest the easiest way to restore resilience would be to protect tree regeneration from the high herbivore pressure in the area.


If you want to read more about our study you can find the paper here and details of our project on forest resilience can be found here. Oh, and here’s a post I wrote about my project a while back. Also, feel free to comment below!

Tropical deforestation causes dramatic biotic homogenisation

Although species richness is most ecologists go-to metric to ‘take the temperature’ of an ecosystem, it is not always the most useful. Even when species richness doesn’t change much over time many species may be being added to or lost from a community. Changes in human land use can cause loss of a particular taxonomic or functional groups, which can have important implications for ecosystem processes such as pollination or seed dispersal. This non-random loss of species as a result of human impacts can result in biotic homogenisation – where the communities in different location become more similar to each other. Biotic homogenisation has been seen all over the world in response to drivers like urbanisation, agricultural land-use change, and eutrophication.

However, up until recently, there had been little work on how biotic homogenisation impacted multiple taxonomic groups across landscapes. Work has also been almost entirely carried out at a single spatial scale. Given that taxonomic groups are likely to differ in their response to disturbances and that landscape scale processes may play a critical role in species persistence. Fortunately last week a paper was published by Ricardo (aka Bob) Solar and colleagues in Ecology Letters that attempted to fill these knowledge gaps.

Specifically the paper attempted to determine how much of the change in community composition as a result of changes in tropical forest land-use change were attributable to replacement of species (termed turnover) and loss of species (termed nestedness). Bob and his colleagues did this for birds, dung beetles, plants, orchid bees and ants at 335 sites (!) in 36 different landscapes in 2 regions of Brazil. The sites used were either primary forest experiencing varying degrees of human disturbance, secondary forests, cattle pasture or arable farmland.

In short the paper shows that:

  • Species richness decreases as land-use intensity increases
  • Differences in community composition between deforested sites were much lower than for forested areas
  • Species turnover caused the majority of changes in community composition, but loss of species became more important as the intensity of disturbance increased
Bob_Solar_Fig5
The importance of loss of species (nestedness) in biotic homogenisation increased as intensity of disturbance increased at both (a) local and (b) landscape scales. Taken from Solar et al. 2015.

For me, the most interesting message of the paper the changes in community composition were largely attributable to replacement of species. This suggests that as species are lost following disturbance, colonisation of generalist species initially causes relatively little change in species richness. However, as land-use intensity increases the contribution of species loss to alteration in community composition became more important suggesting that communities in these locations tend to be made up of generalist species that are tolerant to human disturbances.

Conversion of forest to agricultural use led to much greater biotic homogenisation than degradation.
Conversion of forest to agricultural use led to much greater biotic homogenisation than degradation. Photo courtesy of Bob Solar.

Interestingly, the paper also shows that provided that forest cover is maintained there was relatively little biotic homogenisation. So while it is obvious from previous work that the maintenance of undisturbed forests is vital to conserve tropical forest biodiversity, it is also obvious that degraded forest can play an important role in conservation.  This is especially true where few undisturbed forests still exist or degraded forest is widespread such as in SE Asia and Central America.

This work effectively shows that taxonomic homogenisation is occurring at multiple scales as a result of human land-use change. The next step is to see what types of species are being lost/retained. This means looking at the interaction between species traits and the land-use gradient (see more on that here). Previous work has suggested that body size and feeding preferences may play an important role in determining whether bird species can persist in degraded forests. Looking at this will allow us to gain a greater understanding of how biodiversity change may alter ecosystem processes and ultimately the ecosystem services on which we all depend.

Does reduced impact logging in tropical forests benefit carbon storage and species richness?

After a bit of a traumatic review process* we have just had a paper published in Forest Ecology and Management on the impacts of tropical selective logging on carbon storage and tree species richness. I’m really pleased that we finally got this work out there. If you want to give it a look you can get it here.

Selective logging is one of the most widespread drivers of tropical forest degradation. As I have said before around 400 million hectares of tropical forest are now used for logging – an area twice the size of Russia. Or one hundred and ninety two and a half times the size of Wales – if that’s your thing**.

High intensity logging can result in loss of animal species richness, but on the whole logging is seen as one of the least damaging human uses of tropical forests. That said, there are still concerns about its sustainability in the long-term. Poorly managed concessions commonly remove high timber volumes and do not leave enough time between logging cycles to allow forests to recover.

To improve the sustainability of the practice, reduced impact logging has been proposed. This method aims to reduce negative environmental impacts by cutting lianas and vines before logging, identifying which trees to cut and mapping them before logging starts, planning the roads to be built, and training staff in methods to reduce damage to the forest.  The first papers testing this method showed promising results, appearing to indicate that reduced impact logging causes lower carbon emissions when compared to conventional methods.

However, many papers that have  looked at the impacts of reduced impact logging failed to account for the volume of wood taken out of forests. Crucially, if this differs between reduced impact and conventionally logged sites this represents a hidden treatment, which if not accounted for can lead to faulty conclusions. Given that there are calls to pay people who use reduced impact logging as a means to reduce carbon emissions, we need good, solid science to support this policy.

So, we tried to solve the question of whether reduced impact logging still reduces negative effects on residual tree damage, aboveground biomass, and tree species richness using meta-analysis. We compiled data from all over the globe, all from previously published papers.

Locations of study sites where data we used was collected
Locations of study sites where data we used was collected

Cutting to the chase, the results for reduced impact logging were mixed.

It seemed to reduce the damage to residual trees once logging volume was accounted for…

Prop_damaged_vol
Reduced impact logging (blue) tended to cause less residual damage than conventional logging (red) once logging intensity was accounted for

… however, this did not obviously result in reduced biomass losses, and evidence of an effect on tree species richness was poor as well.

AGB_Richness_volume
Effects of logging intensity on (a) aboveground biomass and (b) tree species richness. Reduced impact logging sites are blue points, and conventional sites are red. Note the relatively low intensity for most reduced impact logging sites.

Though residual damage to trees was reduced, this didn’t cause a  reduction in overall biomass loss. This may be the result of a few different factors. Firstly, residual damage is often to smaller trees so it is not necessarily that surprising that this had little effect on biomass. Secondly, we are really lacking enough data to be sure of the relationship between biomass changes and reduced impact logging. Nearly all of the data is from forests logged at low intensity so we cannot say if the slope of the relationship differs from that of conventional logging.

In the case of tree  species richness, the relative lack of change over a gradient of logging intensity is not too surprising. Newly logged areas richness is probably enhanced by fast growing pioneer species. However, richness is not a fantastically useful measure of biodiversity, in the future it would be much more useful to be able to say what type of species are being lost/gained not just the total number of species in a site (see the recent paper by Zuzana Burivalova and colleagues that tries to do this with bird species).

So what does this all mean? Does our study mean that reduced impact logging doesn’t work? The short answer is no. The long answer is a bit more complicated than that.

First we need to decide whether reduced impact logging is synonymous with low yield logging. If it is then that is fine, but we need to be upfront about this. Logging is after all mainly about timber production. However,some people have previously argued that reduced impact logging can reduce damage whilst maintaining yields. If this is true, it would represent a win-win situation.

If we decide that reduced impact logging isn’t synonymous with low yields then our research question needs to change from “Does reduced impact logging cause less damage than conventional logging?” to “How do the impacts of reduced impact and conventional logging vary over a gradient of timber yields?” Generally in ecology we focus too much on using categorical x variables in statistical tests, and this case is a great example of why this approach can hold back our science (see the fantastic post by Brian McGill on this subject here).

Previous studies show that animal species richness declines with increasing logging intensity and reduced impact logging causes lower losses of animal populations. As a result, a combination of reduced impact logging and reduced logging intensity may appear the best way to reduce carbon emissions and biodiversity loss from logging. However, reducing local yields may cause expansion of logging into previously unlogged areas. This mirrors the current land sharing/sparing debate on how to balance agricultural yields and food production. This debate is taking off regarding logging, and I am keen to see more work on tropical logging that acknowledges the importance of yields. As I said to someone at a conference recently, if we ignore the importance of logging yields why study logged forests?

However, to inform this debate we need more powerful tests of different logging methods than we could do in our paper. One possible source of data for this are studies where logging intensity has been calculated for each sample plot used. For most of the studies I used logging intensity was only available at the site level. Getting this detail would give more statistical power to our tests and provide a more solid evidence base for management of tropical forests. Large-collaborative projects such as the tropical managed forests observatory represent a great chance to answer this question in a more satisfactory manner.

*I will write more about this next week.None of the journals were to blame, just some very biased reviewers.

**If any US citizens want this calculating as relative to Rhode island, I did it. It’s 1273.8 Rhode islands.