Predators and livestock – “stayin’ alive.”

President Donald Trump was elected on a platform that included building a great wall whose purpose was to keep out unwanted intruders from the south, and that would be paid for (apparently magically) by these same intruders.  The idea of building a great wall has been around for a long time; the Great Wall of China was constructed over a time period of almost two thousand years to keep out unwanted intruders (this time from the north). Not surprisingly, the cost of that Great Wall was not borne by the unwanted intruders. More recently, in the 1880s, the government of Australia constructed a 5500 km fence designed to keep unwanted dingoes away from sheep that pasture in southeastern Australia. As Lily van Eeden describes, the Australian government spends about $10 million dollars per year to maintain the fence but there are almost no data to compare livestock losses on either side of the fence. Thus she and her colleagues decided to look at what was being done globally to evaluate the effectiveness of different methods of protecting livestock.

DingoFencePeter Woodard

The Dingo fence across southeastern Australia. Credit Peter Woodard.

The researchers grouped livestock protection approaches into five different categories: lethal control, livestock guardian animals such as dogs, llamas and alpacas, fencing, shepherding and deterrents. Lethal control includes using poison baits and systematic culling of populations of top predators. Deterrents include aversive conditioning of problem predators, chemical, auditory or visual repellents, and protection devices such as livestock protection collars.

Screen Shot 2018-01-23 at 10.28.16 AM

A guardian dog emerges from the midst of its flock in Bulgaria. Credit: Sider Sedefchev.

Van Eeden and her colleagues then did a meta-analysis to see which approach worked best. You can check out my blog from Aug. 2, 2017 (“Meta-analysis measures multiple mycorrhizal benefits to plants”) for a more detailed discussion of meta-analyses. Very briefly a meta-analysis is a systematic analysis of data collected by many other researchers. This is challenging because each study uses slightly different techniques and has different levels of rigor. For this meta-analysis, van Eeden and her colleagues used only two types of studies. One type is a before/after design, in which researchers kept data on livestock loss before the mitigation treatment as well as after. The second type is a control-impact design, in which there was a control group set aside, which did not receive the mitigation treatment. Each study also needed sample sizes (number of herds and/or number of years), means and standard deviations, and had to be run for at least two months to be used in the meta-analysis.

The researchers searched several databases (Web of Science, SCOPUS and European Commission LIFE project), Google Scholar, and also used more informal sources, to collect a total of more than 3300 records. However, after imposing the requirements for types of experimental design and data output, only 40 studies remained for the meta-analysis. Based on these data, all five mitigation approaches reduced predation on livestock. The effect size in the figure below compares livestock loss with the treatment to livestock loss without the treatment, so that a negative value indicates that the treatment is associated with reduced livestock loss. The researchers conclude that all five approaches are somewhat effective, but the large confidence intervals (the whiskers in the graph) make it difficult to unequivocally recommend one approach over another. The effectiveness of lethal control was particularly variable (hence the huge confidence interval), as three studies showed an increase in livestock loss associated with lethal control.

van EedenFig2

Mean effect size (Hedges’ d) and confidence intervals for five methods used to mitigate conflict between predators and livestock.  More negative effect size indicates a more effective treatment. Numbers in parentheses are number of studies used for calculating mean effect size.

Finding that non-lethal management is as effective (or possibly more effective) than lethal control tells us that we should probably be very careful about intentionally killing large carnivores, since, in addition to being cool animals that deserve a right to exist, they also perform some important ecosystem services. For example, in Australia, there are probably more dingoes northwest of the fence than there are south of the fence, so exclusion may  be working. However there is some evidence that there are also more kangaroos and rabbits south of the fence, which could be an unintended consequence of fewer predatory dingoes. Kangaroos and rabbits eat lots of grass, so keeping dingoes away could ultimately be harming the sheep populations. Dingoes may also kill or compete with invasive foxes and feral cats, which have both been shown to drive native species to extinction, so excluding dingoes may increase foxes and cats, threatening native species.  Van Eeden and her colleagues argue that different mitigation approaches work in different contexts, but that we desperately need evidence in the form of standardized evaluative studies to understand which approach is most suitable in a particular context.

van Eeden Fig.3

Context-specific approach to managing the co-exstence of predators and livestock.

In all contexts, cultural and economic factors interact in mitigating conflict between humans and carnivores. The dingo is officially labeled as a wild dog, which invaded Australia relatively recently (about 4000 years ago), so the public perception is that this species has a limited historical role. Other cultures may have a different view of their predators. For example, the Lion Guardian project in Kenya, which trains and supports community members to protect lions, has successfully built tolerance for lions by incorporating Maasai community cultural values and belief systems.

To use a phrase that President Trump recently forbade the Centers for Disease Control to use in their reports, our decisions about predator mitigation should be “evidence-based.” We need more controlled studies that address the success of different mitigation approaches in particular contexts. We also must understand the costs of removing predators from an ecosystem, as predator removal can initiate a cascade of unintended consequences.

note: the paper that describes this research is from the journal Conservation Biology. The reference is van Eeden, L. M., Crowther, M. S., Dickman, C. R., Macdonald, D. W., Ripple, W. J., Ritchie, E. G. and Newsome, T. M. (2018), Managing conflict between large carnivores and livestock. Conservation Biology, 32: 26–34. doi:10.1111/cobi.12959. Thanks to the Society for Conservation Biology for allowing me to use figures from the paper. Copyright © 2018 by the Society for Conservation Biology. All rights reserved.

Cat and fox: agents of Australian extinctions

Australia’s drylands are famous for their assemblage of ultra-cool mammals. As one example, it is difficult for us non-Australians to imagine a more endearing creature than the rock-wallaby pictured below.

Black-foted Rock-wallaby

Black-footed rock wallaby. Credit: Peter McDonald.

Unfortunately, numerous species of Australia’s dryland mammals are going extinct. Many of these extinct species weigh between 35 and 5500 grams – a weight range that researchers have described as the critical weight range (CWR). Peter McDonald and his colleagues wanted to know what was causing these extinctions, and why were they most prevalent in the CWR. They considered two hypotheses. First, perhaps the land was becoming less productive, either from habitat destruction by humans, or as a result of changing climate. Reduced plant abundance could cause herbivorous mammals to go extinct. An alternative hypothesis is that perhaps newly introduced predators, notably feral cats and red foxes, were killing the native mammals so effectively, that they were disappearing from the Ausralian drylands.

Previous research indicated that extinction rates were lower in areas that had more species living in trees and around rocks, leading McDonald to think that maybe habitat was influencing extinctions in important ways. In particular, he realized that rugged mountainous areas might have fewer predacious cats and foxes, and secondly that these two predators tend to go for prey within the CWR. Putting these ideas together, perhaps mountainous areas are refuges for Australia’s dryland CWR species, protecting them from predator-driven extinction. If so, mammal species richness would be highest in rugged, protected areas, and lowest in more open areas. If, on the other hand, mammals are going extinct because overall productivity is declining, we would expect overall species richness to be greatest in the most productive areas.
McDonald and his colleagues tested these two competing hypotheses by censusing mammals in four different types of habitats in Tjoritja National Park within the MacDonnell Range of central Australia. These were (1) mountain areas dominated by a sparse assemblage of shrubs and clumps of spinifex grass, (2) spinifex grasslands (with a more abundant cover of spinifex than found in the mountains), (3) Acacia shrublands, and (4) alluvial woodlands, which were most productive with richest soils.



Mountain refuge habitat_PeterMcDonald

Mountains. Credit: Peter McDonald



Spinifex grasslands


Acacia shrubland


Alluvial woodland

The researchers set up a variety of different mammal traps in 90 different sites representing these four habitats to capture and identify small mammals, and they detected larger mammals by searching for fresh scat at each site. The researchers estimated productivity with the normalized difference vegetation index (NDVI), which uses satellite imagery to measure the green-ness, and hence productivity, of a site or region.

In support of the predation hypothesis, more mammal species were found in the most rugged terrain.


Number of mammal species per site in relation to ruggedness of terrain. The curve is the fitted value of the regression equation.  The shaded area represents the 95% confidence interval.

In contrast to the productivity hypotheses, fewer mammal species were found in the most productive sites


Number of mammal species per site in relation to productivity of terrain as measured by the NDVI. The curve is the fitted value of the regression equation.  The shaded area represents the 95% confidence interval.

While it’s useful to evaluate both hypotheses by measuring current species richness, the researchers also needed to know how many species were actually driven to extinction in the time since cats and foxes invaded. They reconstructed historic species richness for each habitat based on subfossil remains (remains of organisms that are only partially fossilized), from indigenous knowledge supplied by aboriginal Australians, and from historical accounts in the early literature.

They discovered that CWR extinctions were most prevalent in alluvial (12/12 species) and acacia (7/7 species) habitats. Spinifex habitas lost 5/6 CWR species, while mountainous habitats only lost 2/6 CWR species. Importantly, species outside of the CWR have survived relatively well in all habitats, further implicating cats and foxes as the agents of extinctions.


Current (extant) and historic (pre-invasion by cats and foxes) mammalian species richness in the four habitats. The dots are the mean weight, and the lines are the weight ranges for each species.  The shaded area represents the critical weight range (CWR)

More support for the the predation-habitat link comes from recent research that indicates that red foxes are absent from the mountain habitat, while feral cats are substantially less abundant. Even when present, cats are much less efficient hunters in the mountain habitat because the complex rock structure affords more refuges to prey items.

Feral cat with fat-tailed antechinus_NTG

Feral cat captured on camera with a fat-tailed Antechinus. Credit Tony Griffiths.

Across Australia, many CWR species have gone extinct in regions colonized by cats and foxes. McDonald and his colleagues provide solid evidence that these introduced predators are responsible for these extinctions. They urge researchers to explore other mountainous regions in Australia to see if they too are acting as refuges for CWR mammals.

note: the paper that describes this research is from the journal Conservation Biology. The reference is McDonald, P. J., Nano, C. E. M., Ward, S. J., Stewart, A., Pavey, C. R., Luck, G. W. and Dickman, C. R. (2017), Habitat as a mediator of mesopredator-driven mammal extinction. Conservation Biology, 31: 1183–1191. doi:10.1111/cobi.12905. doi:10.1111/cobi.12908. Thanks to the Society for Conservation Biology for allowing me to use figures from the paper. Copyright © 2017 by the Society for Conservation Biology. All rights reserved.

Seagrass scourge: when nutrient enrichment reaches the tipping point

Sean Connell has watched as south Australia has lost vast expanses of kelp forest and seagrasses over the past years. One of the primary culprits associated with loss of seagrass meadows is excessive nutrients, particularly nitrogen, which enters the ecosystem with runoff, and causes an increase in algal epiphytes (epiphytes are small plants that grow on other plants). Epiphytes can negatively affect seagrass by blocking sunlight needed for photosynthesis, and indirectly, by increasing the rate of cellular respiration within the ecosystem, thus using up oxygen needed by seagrass for metabolic processes.


Two dolphins swim above a bed of seagrass off the south Australian coast.

Connell and his colleagues noticed that seagrass loss was often sudden; a large seagrass meadow would appear to be in good shape, and then it would abruptly disappear. They suggested that there might be a threshold effect in nutrient levels that seagrasses can tolerate; that these systems function well until a certain threshold in nutrient levels is crossed, above which there is an abrupt loss of seagrasses. They tested this hypothesis by subjecting plots of the seagrass Amphibolis antarctica to seven different concentrations of dissolved inorganic nitrogen (DIN) over a 10 month period, and monitored the abundance of epiphytes and seagrass over that timespan.

The meadows were about two km offshore from Lady Bay, Fleurieu Penninsula, Australia, in about 5 meters of water. Different amounts of nitrogen fertilizer were wrapped in nylon bags (for slow continuous release of DIN) and staked to the ocean floor. Amphibolis antarctica grows by producing new leaves at the top of each leaf cluster, but at the same time it drops old leaves. Leaf turnover, the researchers’ measure of growth, is simply new leaf production minus old leaf drop. The researchers tied on a small nylon cable at known locations on selected plants, noted how many leaves were above and below each tie at the beginning of the experiment, and recounted leaf number 10 months later. Finally, the researchers measured epiphyte growth by microscopically viewing a sample of seagrass leaves, and counting the number seagrass leaf cells that were covered by epiphytes.

Seagrass growth was relatively unaffected by all tested DIN levels.


Leaf production per day in relation to concentration of DIN.

However, leaf drop showed a strong threshold effect; leaf drop rates increased sharply between 0.13 – 0.15 mg/L of DIN.


Leaf drop per day in relation to concentration of DIN.

Putting these two graphs together, you can see (below) that leaf turnover switched from positive to negative at 0.13 – 0.15 mg/L of DIN. Negative leaf turnover translates to a sudden loss of seagrass at that threshold. At least in this system, at this location, 0.13 – 0.15 mg/L of DIN is the tipping point, beyond which the seagrass system suddenly goes into decline.


Leaf turnover per day (left y-axis and red data), and Epiphyte cover (% – right y-axis and green data), in relation to concentration of dissolved inorganic nitrogen.

The graph also shows that the tipping point coincides with an epiphyte cover of approximately 60%. It is possible that increased epiphyte cover may reduce seagrass photosynthetic rates (particularly in lower leaves), so that leaf turnover suddenly shifts into the negative zone, but the study was not designed to identify the underlying mechanism.

Seagrass meadows perform important ecosystem services, such as absorbing excess nutrients from the sediment, and providing habitat and food for a diverse group of grazers and indirectly, for their consumers. Thus seagrass conservation is vital. The danger here is that moderate levels of nutrients do not appear to have much effect on seagrass populations, but there is an abrupt shift to seagrass loss once the nutrient threshold is crossed. This makes the system very difficult to manage, because the loss occurs without warning. Australian ecologists have repeatedly failed to restore lost seagrass meadows, as simply reducing nutrient levels does not reverse the process. Thus anticipating seagrass loss before it happens is the most viable management solution for this critical ecosystem.

note: the paper that describes this research is from the journal Conservation Biology. The reference is Connell, S. D., Fernandes, M., Burnell, O. W., Doubleday, Z. A., Griffin, K. J., Irving, A. D., Leung, J. Y.S., Owen, S., Russell, B. D. and Falkenberg, L. J. (2017), Testing for thresholds of ecosystem collapse in seagrass meadows. Conservation Biology, 31: 1196–1201. doi:10.1111/cobi.12951. Thanks to the Society for Conservation Biology for allowing me to use figures from the paper. Copyright © 2017 by the Society for Conservation Biology. All rights reserved.

Frogs face fatal fungal foes

Pathogens are organisms that cause disease, and like all organisms, they obey evolutionary principles. Pathogens that survive and reproduce successfully in a particular environment will have more offspring than those that are less successful, thereby passing on those traits that promote successful reproduction to future generations. The problem is that many pathogens change their environment in a way that makes their environment less hospitable for their own survival or reproduction. For example, the fungal pathogen Batrachochytrium dendrobatidis (Bd) causes chytridiomycosis in its amphibian host, which may severely reduce the host population size to the point where few individuals survive. If the host population goes extinct, then there are no hosts for the fungal offspring to infect.

Scheele fungal spore

Scanning electron micrograph of Batrachochytrium denbdrobatidis spore. Credit: Dr. Alex Hyatt, CSIRO Livestock Industries’ Australian Animal Health Laboratory.

Fortunately for Bd, but unfortunately for amphibians, there are several ways out of this conundrum. One approach is a reduction in pathogenicity so that a pathogen’s host species is able to tolerate the infection (and of course, natural selection will at the same time favor an increase in the host species’ tolerance for the pathogen). A second approach is to broadcast a wide net by infecting many different species. That way if one host species goes extinct, there are always many other species to infect. Bd infects over 500 species of amphibians, and has been implicated in the extinction of over 100 amphibian species, and the severe decline of an additional 100 species.

Ben Scheele and his colleagues wanted to know why the endangered northern corroboree frog, Pseudophryne pengilleyi, was declining in southeastern Australia. Several previous studies showed that many corroboree frog populations declined or went extinct in that region over the past 20 years, while the abundant common eastern froglet, Crinia signifera, showed no signs of decline over the same time period. Pilot studies showed that eastern froglets were heavily and commonly infected with Bd. The researchers reasoned that eastern froglets could be acting as a reservoir for Bd, so that corroboree frog populations are being decimated by association with Bd-infected eastern froglets.

Female Ppen copy Hunter

Female Pseudophryne pengilleyi. Credit: David Hunter.

Preliminary surveys indicated that the decline of corroboree frogs was not uniform across the study site; in fact there were some newly discovered populations that were doing very well. The researchers defined three types of sites in their research area. Absent sites (40 in total) had corroboree frogs in 1998, but the population went extinct by 2012. Declined sites (17 in total) had a greater than 80% decrease in abundance since 2000. New sites (25 in total) were newly discovered since 2012, and had much higher population densities than declined sites.


Study area in southeastern Australia, showing locations of Absent, Declined and New sites.

Unfortunately, it is impossible to visually distinguish an infected frog from an uninfected frog, at least until the few hours before death. But the researchers needed to be able to tell if a frog had chytridiomycosis. So they collected skin swabs from the frogs during the breeding season – only working at night to ensure cool humid conditions which minimized frog stress. They then did real time PCR on these samples to quantify the intensity of Bd infection.

Scheele and his colleagues had three important questions they were now prepared to answer. First, how prevalent is Bd in these two species? They found that infection rate was much higher in eastern froglets (79.4%) than in corroboree frogs (27.3%). The intensity of infection (measured by the number of fungal spores) was also much greater in eastern froglets than in corroboree frogs.

Second, do eastern froglets act as a reservoir for Bd, leading to infection and decline of corroboree frog populations? As we discussed earlier, the two species coexist at some sites, but not at others. If eastern froglets act as a reservoir for Bd, we would expect corroboree frogs to have higher infection rates at sites they share with eastern froglets, than they do at sites without eastern froglets. In support of this prediction, Bd prevalence in corroboree frogs was 41.4% at sites with eastern froglets, but only 2.6% at sites with no eastern froglets.

crinia and pengilleyi 3

C. signifera (left) and P. pengilleyi spending quality time together in a P. pengilleyi nest. Credit: David Hunter.

Finally, the researchers want to identify conditions that will promote corroboree frog recovery. They approached this quantitatively by modeling the probability of a site being classified as Absent, Declined or New, in relation to eastern froglet abundance. Based on their survey data of 81 sites, those sites with the highest eastern froglet abundance are most likely to be classified as Absent (corroboree frog extinction), while sites with very few eastern froglets are most likely to be classified as New (thriving corroboree frog populations).


Probability of a site being classified as Absent, Declined or New, based on eastern froglet abundance. Data are log transformed. Dashed lines are 95% confidence intervals.

Scheele and his colleagues conclude that eastern froglets are a reservoir host for Bd, and have played a major role in the decline in corroboree frog populations. The researchers point out that, in general, areas lacking reservoir hosts may provide endangered species with refugia from infectious disease. For managing endangered species, conservation biologists should carefully monitor sites for the presence of reservoir hosts so they don’t reintroduce rare and endangered animals into locations where they will be attacked and killed by pathogens.

note: the paper that describes this research is from the journal Conservation Biology. The reference is Scheele, Ben C., David A. Hunter, Laura A. Brannelly, Lee F. Skerratt, and Don A. Driscoll. “Reservoir‐host amplification of disease impact in an endangered amphibian.” Conservation Biology 31, no. 3 (2017): 592-600. Thanks to the Society for Conservation Biology for allowing me to use figures from the paper. Copyright © 2017 by the Society for Conservation Biology. All rights reserved.

Metallic starlings: a rain of terror

I am a slow learner. Several times in the past few years I have paddled my canoe under a particular sycamore tree in the New River in Radford, Virginia. Each time I do so, I am greeted by large numbers of cormorant poop bombs dropped by the dozens of cormorants that seem to find that particular tree to their liking, and this particular canoeist not to their liking. Fortunately, cormorants have bad aim, but unfortunately it is not that bad.

Daniel Natusch and three other researchers wanted to know how an analogous form of nutrient enrichment from large colonies of nesting Metallic Starlings (Aplonis metallica) affects the nearby ecosystem in a tropical Australian rainforest. They were interested in this question because it was obvious that the ground below the nesting colony trees was basically devoid of vegetation; they describe it as “an open moonscape”, contrasting sharply with the thick rainforest nearby. Other studies have shown that nutrient enrichment from bird guano leads to increased vegetation density – so why is this ecosystem different?


Lockerbie Scrub rainforest, in Cape York Peninsula, Australia, showing colony tree with dead zone (left) and a continuous rainforest (right)



Dan Natusch conducts herpetological research with his son Huxley. Credit: Jessica Lyons













The researchers compared the biological, chemical and physical environment underneath 27 different colony trees to the environment underneath a randomly chosen tree 100-200 meters from the colony tree. As expected, they found very little vegetation near colony trees, in contrast to relatively dense vegetation near the randomly chosen trees.




Vegetation cover (left) and number of live stems (right) in relation to distance from the colony or randomly chosen tree (Point 0 on X-axis).  Negative numbers are downslope and positive numbers are upslope from the tree.


Soil analyses showed that the soils under the colony trees had much higher concentrations of important nutrients. For example, phosphorus levels were more than 30 times greater, and ammonium nitrogen was about four times greater under colony trees than under the randomly chosen trees. The researchers wondered whether these nutrient levels were so high that they were toxic to vegetation. That would account for the dead zone under the colony trees. An alternative hypothesis is that animals (pigs and turkeys in particular) may be attracted to these high nutrient areas under the colonies, and may either kill germinating plants by eating or trampling them.

Feral pigs (Sus scrofa) rooting and trampling under a colony tree. Credit Daniel Natusch

To test both hypotheses, at the beginning of the breeding season the researchers covered a portion of the colony tree region with metal cages (exclosures) that prevented turkeys and pigs from gaining access. They discovered a much greater number of seedlings under the exclosures in comparison to the areas where turkeys and pigs could access the seedlings.

natuschfig7They concluded that nutrient levels were not toxic to seedlings, but that pigs and turkeys were either eating or trampling the seedlings as they emerge. As you can see, the number of exclosure seedlings dropped sharply in July, in part because rainfall declines sharply in June, which leads to high plant mortality, particularly in the unshaded dead zone. But in addition, feral pigs broke into all of the exclosures that summer to access the seedlings and the nutrient-rich soil.

Do these dead zones actually benefit the starlings in any way? One possible advantage is that dead zones prevent snakes from climbing nearby trees and vines to gain access to the nests that are located high in the canopy of the colony tree. However there is good evidence that colony trees suffer high mortality, as 10 of the 27 colony trees died within three years of the study. Trees that fall during the nesting period could lead to the failure of all of the nests within that colony tree.



A scrub python (Morelia amethistina) puts the squeeze on a juvenile Metallic Starling. Credit Daniel Natusch.


Why do we find dead zones beneath colonies of Metallic Starlings, and increased plant growth rate, larger plant size and greater plant diversity beneath the colonies of several other bird colonies? Most previous studies have looked at sea-bird colonies on small islands that have few terrestrial herbivores, so germinating seedlings are relatively undisturbed. This study occurred in a continuous forest in tropical Australia, which harbored a large population of hungry herbivores. These contrasting findings show the important role of environmental context for understanding how ecological interactions will play out. Given that we humans are continually adding nutrients to our environment (through natural bodily function and when we fertilize our fields), we need to carefully consider the biotic and abiotic players in the ecosystem, so we can predict the effects we are having on the environment.

note: the paper that describes this research is from the journal Ecology. The reference is Natusch, D. J. D., Lyons, J. A., Brown, G. P., & Shine, R. (2017). Biotic interactions mediate the influence of bird colonies on vegetation and soil chemistry at aggregation sites. Ecology 98(2): 382-392. Thanks to the Ecological Society of America for allowing me to use figures from the paper. Copyright © 2017 by the Ecological Society of America. All rights reserved.