ace-lab

[ace-lab]

Why did you count so many butterflies? Insights on sampling conditions and Pollard walk counts

Humans like to count things. Indeed, one of the favourite activities for ecologists is wandering in nature and counting species: we learn quickly that some critters are more abundant in certain habitats and seasons, and often use this information for locating, protecting, and managing biodiversity across the landscape. 

When we sample butterflies (or any other organisms - including plants! (Dennet and Nielsen 2019), the number of individuals counted depends on our ability to detect a species. For instance, a certain trap type might be more attractive for some species, but the fact that those species represent the most conspicuous samples in our trap doesn't necessarily mean that they are also the most common in the area. Similarly, if we are counting butterflies along a transect ("Pollard Walks", from Ernie Pollard's famous monitoring scheme https://www.ceh.ac.uk/news-and-media/blogs/40-years-butterfly-monitoring-celebration), some species are easier to observe than others, and this affects our understanding of abundance patterns within a community of species.

This phenomenon is formally defined as "detection probability" (or "detectability"). Focusing on butterflies and Pollard walks, detectability affects comparisons both between and within species, and it depends on several factors including (i) the characteristics of a species (e.g., it's easier to detect large, colorful butterflies that fly for longer periods of time), (ii) the characteristics of the site where samples are conducted (e.g., it's harder to see the same species in a dense forest than in a meadow), and (iii) the conditions at which samples are conducted (e.g., the same transect conducted at noon vs. midnight in the same day will likely result in different counts, even though the number of butterflies in that site is arguably the same).

These factors are all self-evident per se, but are intertwined in determining how many butterflies we count when we are sampling butterflies in the field, and can be complex to disentangle. Pragmatically, it is important to understand if we count more butterflies because there are actually more butterflies, or because the conditions at which samples are conducted favour one species or one site. Indeed, for populations of rare species, we want to protect places that actually harbor more butterflies - not places where fewer butterflies are easier to see.

In our recent paper (available open-access at https://esajournals.onlinelibrary.wiley.com/doi/full/10.1002/ecs2.3101) we asked whether within "recommended sampling conditions" our ability to detect butterflies was constant. This is a common assumption in butterfly studies, but we demonstrate that this is not always the case: even within standardized conditions (10 AM - 4 PM, temperature > 17 degrees celsius, wind speed < 25 km/h), our ability to detect butterflies varied between ~ 5% and 50% depending on hour of the day, temperature, cloud cover and wind speed. In practice, this suggests that rare species are likely often missing in our samples even if present at a site, particularly when we're not sampling in optimal conditions.

Recognizing these limitations and documenting the conditions at which samples occurred is a first step to properly interpret what we see in the field - and, thus, to effectively protect and manage biodiversity. Accounting for detection probability is therefore a powerful approach when monitoring and assessing changes in biodiversity. 

[ace-lab]

Mapping landscape rarity in Alberta’s boreal plants

Relative to other systems, the boreal forest is species poor with low rates of endemism. However, many rare and uncommon species occur in northeastern Alberta, and their conservation is dependent on understanding where these plants are likely to be on the landscape. From 2011 to 2015, Dr. Nielsen, in collaboration with the Alberta Biodiversity Monitoring Institute, designed and oversaw a terrestrial vascular plant monitoring program designed to target rare species.

Rare species, as you may expect, are inherently difficult to survey. Knowledge of occurrences, species ecology, and population dynamics are often lacking, particularly in areas like the boreal where survey effort tends to be low due to difficult ground and poor accessibility. Dr. Nielsen used an iterative, model-based approach to sampling, where landscape factors such as terrain and soil pH were used to predict areas that were most, and least, likely to contain rare species. Over four years, we sampled 602 unique plots and found 42 populations of nine rare species. Uncommon species (those ranked at S3) were found at nearly 100% of plots, clearly demonstrating the advantages of model-directed sampling.

The data produced by this project are publicly available and have been used in multiple manuscripts from the ACE Lab and colleagues to date. To read more about the boreal forest, rare plants, and the details of this extensive sampling effort, check out our new, interactive StoryMap, “Landscape Rarity”!

https://uofa.maps.arcgis.com/apps/Cascade/index.html?appid=cf214263b9f447c0abf71b665cc79904

[ace-lab]

Detectability of cryptic graminoid species

Recent work in the ACE lab focused on imperfect detection, meaning the failure to observe species where they are present (link: http://ace-lab.tumblr.com/post/172295897598/investigating-detection-success-lessons-from). We are back addressing this topic again in a new publication focused on a favourite group in the ACE lab – graminoids.

Graminoids are grass-like plants, defined here as the sedge (Cyperaceae), rush (Juncaceae), and grass (Poaceae) families. Evidence in the literature suggests graminoids may be consistently under-detected relative to other types of plants (e.g., forbs), so we took a deep-dive into their detection using sedges (Carex) as a model group. Sedges are an excellent focal genus because they vary widely in form and size in the boreal, and occur in virtually all landcover types.

Top: Short species with small seed heads, such as Carex deflexa, were the most poorly detected in our experiment.

Bottom: Tall species with large seed heads, such as Carex rostrata, were almost never over-looked by the observers used in this study.

We focused on site- and species-specific differences in this experiment, meaning we considered the influence of variables like vegetation density on transects, and species morphology, on detection.  Building on our previous work, we expected that abundance would be the major determinant of success, and we suspected dense sites (those with a complex understory) and small morphology (short, with small seed heads) would both negatively influence detection. We inventoried 50 belt transects (100 x 2 m) with a standardized effort of 0.15 minutes/m2 per observer, with two independent observer surveys per transect. We took an approach of grouping sedge species by their morphology and created six categories based on field measurements and general appearance.

Using pseudoturnover (the difference in observer species lists due to overlooking), we found no evidence of a bias toward graminoids as a group over other life forms. The greatest pseudoturnover values for graminoids were associated with sites having high forb and short shrub cover, where high foliar cover likely obscures graminoids and reduces consistency between observers.

Overall, we showed relatively high detection among the six groups of sedges, ranging from 0.82 for short sedges with small seed heads to 0.99 for tall species with large seed heads. We found that detection failures at a site were best explained by sedge species abundance and morphology, but not supported by site structure variables. In contrast, detection delays (where an observer noted a species after it was first present), were related to abundance and site structure, specifically trees and tall shrubs, but not to morphology. This suggests that the distractions caused by physical obstruction cause observers to be slow to notice species, regardless of their form.

Predicted detection probabilities given Carex species abundance (% cover) within transect segments (20 m2) as per the best supported model. Lines indicate predicted detection probabilities, shaded bands indicate 95% confidence intervals. Species segment abundance is truncated at 10% for ease of viewing, as all groups had predicted detection probabilities ~1 past this value. Note that detection probabilities are poorly estimated at 0% cover.

Predicted detection probabilities given Carex species abundance (% cover) within transect segments (20 m2) as per the best supported model. The black line indicates predicted detection probability, the shaded band indicates a 95% confidence interval. Note that detection probability is poorly estimated at 0% cover.

We demonstrate that detection of species at low abundance, particularly at <5% cover, can be significantly <1, and delays in detection can occur at even higher abundance, up to 25%, even with high survey effort relative to many plant survey practices in Alberta. Using broad morphological groupings may be helpful for future work, where group-based detection probabilities could be used to inform occupancy estimates where detection data are not available. Searches for species with small morphology must employ even greater survey effort than was used here to ensure adequate detection. Graminoids are a fascinating group, despite their challenges in observation and identification in the field, and we look forward to future work on these beautiful species within the ACE lab.

Additional paper summary: https://jvsavsblog.org/2019/02/17/detectability-of-species-of-carex-varies-with-abundance-morphology-and-site-complexity/

Citation:

Dennett, J.M., and S.E. Nielsen. 2019. Detectability of species of Carex varies with abundance, morphology, and site complexity. Journal of Vegetation Science (early view). https://doi.org/10.1111/jvs.12713

[ace-lab]

Finding landscapes of stability and conservation efficiency in the face of climate change

There is no shortage of gloom and doom in climate-change news these days. Without drastic emissions reductions or technological fixes, the planet’s ecosystems will continue to change rapidly, with novel conditions likely challenging our ability to manage and conserve species. So what is a land manager to do? There is no silver bullet. Managers will need to be creative, employ multiple strategies, and be prepared to adapt if things don’t work out. But all of that can be costly. Although we need to acknowledge and anticipate change, a complementary strategy is to find and protect areas of relative stability as refuges for species and ecosystems from climate change. These “refugia” may be considered relatively efficient, low-risk investments—the savings bonds in a conservation portfolio.

In our new paper in Global Ecology and Biogeography, we started with the premise that refugia locations vary by species, and should be considered with respect to individual species’ environmental niches. But we also recognized that common climatic limitations on species distributions should result in high refugia overlap among species. Working at a macro scale, we developed an index to identify multi-species refugia across Canada and the United States. We focused on trees and songbirds, taking advantage of intensive sampling and knowledge for these species, and models of existing conditions and projections under climate change developed at a 10-km resolution for Canada and the USA by the Canadian Forest Service (trees) and the National Audubon Society (songbirds). Applying the concept of climate velocity—the speed that an organism must move to keep pace with climate change—our index identifies not just those areas in which species can persist, but also the climatically suitable areas to which they can reasonably relocate within the next century.

Areas with the highest overall refugia potential for many species were found in steep mountain terrain, where trees and birds can disperse up-slope to cooler conditions, and in coastal regions, where the effects of warming are moderated by ocean influences. Songbird refugia were strongly associated with elevation, while coastal proximity and landform composition (particularly headwaters) were important for both groups. But this is only part of the story. In addition to these topographically unique “universal” refugia, we identified other areas of high refugia potential for particular groups of species. These climatic “stepping stones” to future distributions depended on species’ environmental niches and southern range limits, and were not strongly linked to topography. Additionally, we found that areas of greatest refugia potential were not generally areas of high species diversity, highlighting the complex trade-offs involved with conservation planning under climate change.

Multi-species end-of-century (2071-2100) refugia indices averaged across (a) 324 tree species, (b) 268 songbird species, and (c) all species combined, weighted by projected climate-change response. Legend breaks are defined by percentile values. (d) Difference between refugia percentile ranks for trees and songbirds. Radiative forcing values: Representative Concentration Pathway 8.5.

One of the main goals of the paper was to identify the elements of climate that limit refugia potential in different parts of northern North America. For the highest-value refugia, we found that precipitation was generally most limiting in the north, and warm temperatures and moisture availability were limiting in the south. Tree refugia were more limited by precipitation and moisture, while songbird refugia were more limited by temperature, suggesting differences in the ecological processes that govern warm-end range limits for different taxa in different regions. Our framework can be applied to other regions, species, and time periods to generate and explain biologically meaningful indices of macrorefugia for conservation planning.

Spatial data layers from this work can be found on the AdaptWest website at https://adaptwest.databasin.org/pages/climatic-macrorefugia-for-trees-and-songbirds.

Citation:

Stralberg, D., C. Carroll, C. Wilsey, J. Pedlar, D. McKenney, and S. Nielsen. in press. Macrorefugia for North American trees and songbirds: climatic limiting factors and multi-scale topographic influences. Global Ecology and Biogeography. https://doi.org/10.1111/geb.12731

Diana Stralberg (diana.stralberg@ualberta.ca), 3 April 2018

[ace-lab]

Investigating detection success: lessons from trials using decoy rare plants

Field surveys are often the first action of ecological inquiry. We rely on accurate field data to answer questions about species presence, abundance, distribution, and demography, yet, we also know humans to be imperfect observers. When detection errors are non-random (i.e. scale with some variable of the survey), bias can be introduced, undermining our ability to answer important research questions. When study organisms are mobile the likelihood of imperfect detection and bias are more obvious, but vascular plants may be as poorly detected as cryptic mammal or bird species. In general, plant ecologists are less likely to account for imperfect detection in study design and analysis than those who study mobile taxa, presumably due to an ongoing perception that a static organism will be detected where it is present. Examples of literature focused on imperfect detection in vascular plants are growing in recent years and continue to make clear that the notion of plants as perfectly detectable is impractical in the majority of cases. Understanding what observer characters or survey attributes relate most to detection will allow for greater accuracy and efficacy of field surveys and improve confidence in reported absences of species of conservation concern.

In this paper, we used decoy rare plants (planted individuals of a species not currently growing in the area) and 29 volunteer observers over two trials in separate years to look at vascular plant detection in mixed-wood forests. Our trials were similar to Moore et al. 2011 (Estimating detection-effort curves for plants using search experiments, Ecological Applications) where observers were asked to search for specific species in survey plots but were unaware of their true presence or abundance. Observers were asked to search until they felt that they had adequately covered the plot and were told that none, one, or both species may be present, thus, they would search to saturation or until encountering both target species. Over both trials we considered the influence of observer experience, species appearance and identity, plot size, and abundance and arrangement of the target species.

Examples of decoy plants immediately prior to planting in experimental survey plots. The two species used in Trial One were western willow aster (Symphyotrichum lanceolatum, front left and right) and crowfoot violet (Viola pedatifida, front-center).

Our results suggest that the detection of cryptic species is low (0 – 35%) when plot size is large, and that plant density (# individuals/unit area) is the most important factor determining detection success. Morphologically distinct species, and even those in flower, may still go unnoticed in plots where abundance in low, suggesting that field surveys targeting uncommon species may need significantly higher effort then what is typically employed, particularly in our region of Alberta. We reported variation among observers but no consistent relationship was observed between observer experience and detection success, suggesting that even intermediate botanists may be equally effective as experts under certain search environments for single species. Our results support the use of small plots where possible, given that observers are unlikely to find plants in large search areas, even though some trade-off in area may be required since local size of populations in rare species is likely low thus limiting their encounter rate. Further, few survey plots went entirely undetected when considering all observations, thus, repeat surveys may compensate for low detection probabilities on a per site basis and we encourage this approach to plant survey where accurate detection is essential.

J. Dennett, March 27, 2018

The full publication, Dennett et al. 2018, can be found here

https://link.springer.com/article/10.1007/s11258-018-0819-1?wt_mc=Internal.Event.1.SEM.ArticleAuthorOnlineFirst

Citation:

Dennett, J.M., Gould, A.J., Macdonald, S.E. & Nielsen S.E. 2018. Investigating detection success: lessons from trials using decoy rare plants. Plant Ecology (online first).

[ace-lab]

Brown bear: the quintessential omnivore

Brown bears (Ursus arctos) are pretty interesting animals in that they are carnivores with an omnivorous diet. Around the world, brown bears have been documented eating a wide variety of foods, such as vertebrate and invertebrate prey, grasses and forbs, roots and bulbs, and fruits and nuts. And, as most people know, brown bears also have a fondness for human foods, which often leads to trouble.

It’s easy to imagine then that brown bears might eat indiscriminately. There is evidence to the contrary, however, as studies of captive bears have shown that they in fact mix their diet in such a way as to maintain a preferred ratio of the dietary macronutrients, protein, carbohydrate, and fat. The preferred ratio selected by bears also maximized their mass gain (primarily fat) per unit energy intake, suggesting the functional importance of their evolved dietary preferences — brown bears require a certain amount of body fat for successful hibernation and reproduction.

The story is not so simple, however, as the foods available to wild brown bears that would allow them to consume such a ratio of macronutrients are generally not available until the late-summer and autumn. These are high-fat and high-carbohydrate foods, such as nuts and seeds (fat), and fruit (carbohydrates). One might then predict that brown bears are likely to experience a large amount of variation in dietary macronutrients consumed between seasons within a population, and that bears are more likely to consume an optimal ratio during the autumn.

Furthermore, ecosystems vary in the types of foods available to bears. Thus, one might also predict that there is a large amount of variation in the proportion of dietary macronutrients consumed between geographically distinct populations of brown bears.

Or is there?

Nutritional ecology studies of other animals, such as mountain gorillas (Gorilla beringei), have found that geographically distinct populations consumed very similar proportions of macronutrients in their diets despite consuming different combinations of foods, suggesting active and shared regulation of macronutrient intake (i.e. a narrow “fundamental macronutrient niche”). Other studies, however, such as one investigating the diet of omnivorous wild boar (Sus scrofa), found a wide range of dietary macronutrient intakes between populations (i.e. a wide “fundamental macronutrient niche”), suggesting that their omnivorous diet served to enable them to occupy a wide range of habitats.

In our recent paper, we set out to investigate this by estimating the proportions of macronutrients in the diets of published bear studies around the world. We found that, unlike mountain gorillas and similar to boar, there was a great deal of variation in the proportion of macronutrients consumed by brown bears among populations and seasons (i.e. a wide fundamental macronutrient niche). The diets of brown bears on average were higher in protein and lower in carbohydrate during spring, and conversely higher in carbohydrate and lower in protein during autumn. The average proportion of fat consumed in diets was relatively constant among seasons. As expected, the diets of brown bear populations tended to be closer to the ratio selected by captive bears during the autumn, which is prior to their hibernation period.

Another interesting thing we found was that diets of populations which also consumed human-sourced foods, such as agricultural crops and domestic livestock, were higher on average in the proportion of carbohydrate, and lower in the proportion of protein, compared to populations with natural diets. The proportion of fat consumed also tended to be higher in diets of bears consuming human-sourced foods, but to a lesser extent than carbohydrate. This meant that more of the diets of bears consuming human-sourced foods were closer to the “optimal” ratio during autumn compared to those with natural diets, with the possibility of consuming such a ratio even during summer.

Our analysis suggests that omnivory in brown bears is an evolutionary adaptation which allows them to occupy a diverse range of habitats and tolerate variation in the availability and nutritional composition of food resources. Furthermore, we show that brown bear populations consuming foods from human sources have different proportions of macronutrient in their diets. The biological effects of differing dietary macronutrient proportions on individual brown bears and populations is an important area of future research into their nutritional ecology. This is especially important given the growing human population in bear habitats, and will help inform on the potential impacts of climate change on grizzly bear food resources.

For more information refer to our paper which is available open-access from Ecology and Evolution:

Coogan SCP, Raubenheimer D, Stenhouse GB, Coops NC, Nielsen SE. 2018. Functional macronutritional generalism in a large omnivore, the brown bear. Ecology and Evolution 8:2365-2376.

Link: http://onlinelibrary.wiley.com/doi/10.1002/ece3.3867/full

Dr. Sean C. P. Coogan, 21 February 2018

[ace-lab]

Improbable destinies (or is it?)

Jonathan Losos’ recent book is an interesting read that very well may shape people’s view on natural selection (evolution) in what can be viewed as a counter to some of the famous ideas of evolutionary biologist Stephen J. Gould.

As Losos points out, for far too long one dominant viewpoint in evolution following Stephen J. Gould idea is that if you ‘replay life’s tape again, you’d get a very different outcome’. In other words, life should not repeat itself. There are too many stochastic factors (asteroids and such) that are unpredictable and ultimately fundamental to shaping life based simply on chance events. But what if life tends towards common solutions to repeatable conditions (environments)?

With one earth and one indefinite stream of time, we simply cannot replay life beyond the thought experiment of Gould (an unfalsifiable theory). Or can we?  A classic way ecologists work is to substitute space for time. For instance, comparing islands or an isolated continent (especially Australia) to another content. When doing this, it turns out that life is surprisingly more similar than one would expect from Gould. Losos draws on his own work on islands to illustrate this, as well as the work of biologist Conway Morris and a number of other scientists now doing laboratory evolutionary studies (some supporting and some less supportive of some of these ideas). They generally argue, however, that life can be quite repeatable (predictable) after all. Although niche space (environments) is diverse, ‘solutions’ towards adapting to these environments are often limited due to fundamental evolutionary constraints that natural selection acts on. The result is convergent evolution. Certain strategies work better than others with natural selection ultimately working to ‘find’ a ‘solution’, regardless of the mistakes it makes along the way. The end result is surprisingly similar forms of life in different places despite being quite phylogenetically dissimilar.

One classic example of convergent evolution is the gray wolf (Canis lupus) - Thylacine (Thylacinus cynocephalus) comparison (see below Figure). Gray wolves from the northern hemisphere, Thylacine (also called Tasmanian wolf) from Australia (now extinct), each evolved independently of one another (Thylacines from marsupials), yet each filling a similar ecological role (niche) to a common opportunity (large populations of moderate-sized herbivores are common to all continents, which in itself represents a ‘solution’ to an opportunity). 

One might still go back further in time and still wonder in a thought experiment like Gould about what would happen in long stretches of time such as what would happen if the asteroid didn’t hit 66 million years ago. Would mammals lose out? That may very well be the case, but as Losos points out based on Dale Russell’s work, that doesn’t mean that eventually we wouldn’t have a 'dinosauriod' characterized by a large brain (advantage) requiring a larger, heavier head that would need to be balanced on top of an upright body, which subsequently would discount the need for a tail (Russell showed that over time animals have steadily evolved bigger brains). A dinosauriod that generally looked human-like in form...  And even without asteroids, it’s entirely plausible that dinosaurs were on the way out (well kind of, given that they are our birds today) with endothermic mammals on their way in, but the important part here is not the name of the species or group who eventually dominated larger animal life, but rather the final form of organisms which may qualitatively be similar in character to current life.

In the end, Losos’ book is easy to read with interesting insight and examples. It should be on every biologist’s nightstand. Conway’s work may be a natural follow-up for those interested in more detail and history, but his use of 100s of pages of footnotes may not make it as accessible of a read as that of Losos.

S. Nielsen, January 26, 2018

[ace-lab]

Butterflies whisper subtle truths into the ongoing fragmentation debate: mobility matters.

The theory of island biogeography (IBG) is among the most influential and broadly applied theories in ecology. Specifically, IBG predicts that the number of species (species richness) living on oceanic islands vary as a function of island size and isolation. The simplicity of the logic is appealing. Population size is theorized to increase with island size—more space and resources support greater numbers of individuals. Working backwards through this relationship, it is clear that populations restricted to small islands are more prone to stochastic extinction events and inbreeding depression associated with small population sizes. Subsequent increases in extinction rates for small islands should reduce their overall species richness relative to larger islands. Beyond island size, the relative configuration of islands also plays a critical role in structuring species richness. Islands that are further isolated from sources of species immigration (e.g., other islands or mainland) are less likely to be colonized by new species. Additionally, islands that are further isolated are less likely have dwindling populations of already existing species “rescued” by conspecific immigrants. In its most simplistic sense, IBG predicts that insular species richness increases with island size and decreases with island isolation.  

Although IBG was developed in the context of oceanic islands, it has had profound consequences in how ecologists think about suitable habitat fragments (“islands”) situated in a sea of unsuitable terrestrial habitat. Indeed, a priori parallels between oceanic islands and habitat fragments on terrestrial landscapes suggest that decreasing fragment size and increasing fragment isolation reduce species richness at both the fragment and landscape level. This application of IBG, dubbed the “island effect hypothesis,” is a powerful heuristic, and many early works concluded that fragmentation effects were of great conservation concern. However, recent research suggests that negative relationships between fragmentation and species diversity are artefacts of habitat loss: generally, total habitat area correlates positively with species diversity, and negatively with degree of fragmentation. It may be that smaller fragments contain fewer species simply because of the “sample-area effect”: if you survey bigger areas, you find more individuals, which belong to more species. To better understand the sample-area effect, envision habitat fragments as “targets” of different areas, and species as “darts” of different colours. If darts are tossed randomly at a landscape of targets, larger targets will accumulate more darts (of different colours) than smaller ones. This sample-area effect is the underlying mechanism behind Lenore Fahrig’s (2013) “habitat amount hypothesis,” which predicts that habitat fragmentation does not reduce species richness after controlling for the deleterious effects of habitat loss. Bearing on the ongoing “SLOSS debate,” which addresses whether single large or several small habitat fragments protect more species, Fahrig suggests that several small and single large habitat fragments will support equivalent numbers of species if total area is held constant.  

In this study, we used butterfly assemblages on islands of Lake of the Woods, Ontario, Canada to decouple habitat fragmentation from habitat loss and resolve relationships between habitat fragmentation and species diversity. Lake of the Woods contains some 14,632 islands, which, rather uniquely, represent an ideal system for testing the island effect and habitat amount hypotheses. Of particular interest, habitat boundaries (i.e., island edges) in this system are strictly delimited by water, and butterflies cannot utilize surrounding aquatic habitats at any life stage. This effectively controls for “matrix effects,” whereby the matrix of unsuitable habitat contributes to species diversity. Using SLOSS-based methods to control for total habitat area (an inverse measure of habitat loss), we compared butterfly species richness across both individual islands and sets of islands to infer whether fragmentation reduced species diversity. Furthermore, by differentiating between potential resident and transient butterfly species within individual islands based on occurrences of larval food plant species, we were able to investigate fragmentation effects on both the complete species assemblage and a subset of potential resident (reproducing) species.

When considering the complete species assemblage (both resident and transient butterfly species), habitat fragmentation did not reduce butterfly species diversity in our study system. This result suggests that habitat configuration has little effect on the number of butterfly species persisting on fragmented landscapes, supporting the habitat amount hypothesis. However, butterfly species vary widely in mobility, and are therefore likely to vary widely in their responses to habitat fragmentation. When highly mobile species occurring on islands without their larval food plants were excluded from analyses, island effects on potentially reproducing species became apparent.  Our study shows that differentiating between potentially reproducing species and highly mobile, transient species observed within individual habitat fragments yields critical insight into the negative effects of habitat fragmentation on species diversity.

This research was published on the journal Oecologia, and can be found here:

https://link.springer.com/article/10.1007/s00442-017-4005-2

http://rdcu.be/zT8W

Zachary G. MacDonald, November 29th 2017

Citation:

MacDonald, Z.G., Anderson, I.D., Acorn, J.H. and Nielsen, S.E. (2017). Decoupling habitat fragmentation from habitat loss: butterfly species mobility obscures fragmentation effects in a naturally fragmented landscape of lake islands. Oecologia: 1-17.

[ace-lab]

How does habitat fragmentation affect boreal forests? Insights from butterflies

Biodiversity interacts and responds to processes that occur across multiple scales, such that the patterns that we observe in nature typically emerge from smaller-scale phenomena that are moderated by larger-scale mechanisms. Understanding the balance between these two determinants of biodiversity is fundamental to addressing applied ecological questions, such as assessing how anthropogenic changes to landscapes affect biodiversity.

Considering measures of local and landscape change, we investigated how butterflies respond to widespread, but localized, disturbances associated with extraction of underground oil sands (“in situ” extraction with wells) in Alberta’s boreal forests. In situ extraction involves relatively little loss of forest habitat, generally less than 15% of total forest cover, mostly due to narrow cleared corridors used to locate the underground oil reserve. The extent of these disturbances is, however, large being comparable in area to England. Therefore, understanding how biodiversity responds to these disturbances is critical to preserving the natural dynamics and communities of the boreal biome. To date, little is known on how insects respond to these patterns of forest fragmentation. We addressed this gap for butterflies by investigating how species assemblage changed in response to different types of in situ oil sands disturbances.

To understand the scale at which these disturbances changed the forest environment for butterflies, we compared the butterflies observed in undisturbed forests with those observed in increasingly larger disturbances, from 3-m wide corridors (seismic lines) to 60 × 60 m well pads. Intuitively, removing a few trees won’t likely elicit a response in the butterfly assemblage, while large forest openings would, but what is unknown is whether the forest clearings associated with smaller footprints, such as those from in situ oil sands extraction, were sufficiently large to affect boreal butterflies. We also assessed the scale at which landscape characteristics moderated localized responses to forest disturbances. We did this by measuring landscape characteristics surrounding our sample sites at different spatial scales and identifying the scale and type of measure (diversity, amount or arrangement of habitat) that most explained patterns in butterfly assemblages.

Our study demonstrated that butterfly diversity and abundance increased substantially within disturbances, even in 9-m wide corridors being 2-fold higher in diversity and 5-fold higher in abundance compared with adjacent forests. Interestingly, 3-m wide corridors that are considered to be “low-impact” by industry did not differ in butterfly composition from adjacent forests supporting their low-impact label. This suggests that disturbance starts to affect butterflies somewhere between 3- and 9-m wide gaps, a threshold that is much smaller in scale than the scales previously studied for butterflies on forest disturbances. And finally, the surrounding landscape was found to moderate butterfly responses, but local disturbances were by far the most important factor in explaining butterfly presence and community composition. This study demonstrates that local effects on butterflies occur on a vast area of boreal forests - even if these disturbances are among the smallest investigated in literature. Presumably, similar effects occur on other analogous groups, such as other pollinator insects.

This research was published on the journal Biological Conservation and can be found here:

https://www.sciencedirect.com/science/article/pii/S0006320717306079

Federico Riva, November 10th 2017

Citation:

Riva, F., Acorn, J.H. and Nielsen, S.E. (2018). Localized disturbances from oil sands developments increase butterfly diversity and abundance in Alberta's boreal forests. Biological Conservation 217: 173–180.

[ace-lab]

Complementary food resources of carnivory and frugivory affect local abundance of grizzly bears

Wildlife population dynamics are influenced (ignoring immigration and emigration) by food resource supply representing ‘bottom-up’ processes (individual performance & carrying capacity) and those factors affecting survival via ‘top-down’ processes. Much previous research on grizzly bears (Ursus arctos) have focused on top-down effects, particularly around the issue of human-caused mortality given that they have naturally low fecundity and no natural predator (apex carnivore). When bottom-up factors are considered, it is often to the isolation of a single resource like salmon, ungulates, or berries, yet grizzly bears are highly omnivorous which suggests that it is the complementarity of resources that is important. 

For interior grizzly bear populations that lack access to salmon, two critically important resources are meat and fruit with each offering different essential macronutrients (meat being a source of protein and lipid; fruit being a source of carbohydrates). Mixing the diets of these two food sources result in more optimal nutrition in bears and thus presumably growth & reproduction of individuals, as well as potentially boosting local carrying capacity. This is especially important to a species whose successful reproduction depends on their body condition prior to denning.  In fact, the species has adapted to unpredictable supplies in food (e.g. berry crop failures) with an evolutionary strategy of delayed implantation of the fertilized egg late in the season prior to denning when the body condition of females are pretty much set for the winter (mating occurs in late spring). 

In this paper, we examined the relative importance of meat and fruit, either individually or combined (additive or multiplicative), in explaining local abundance (density) of grizzly bears in west-central Alberta, Canada. We did this by mapping landscape patterns of digestible energy (kilocalories) of a key fruiting species buffaloberry (Shepherdia canadensis) and ungulate matter (deer, moose, elk & sheep) to the number of bears found at DNA hair snag sites across a 8,624-sq. km region of Alberta, while controlling for the top-down effect on bears using road density. 

Interestingly, complementary resources (fruit + meat) were more supported in predicting abundance than individual patterns of either meat or fruit or even landscape proxies affecting survival of bears (road density). This suggests a nutritionally multi-dimensional bottom-up limitation of grizzly bear populations in Alberta and the need for researchers to consider more broadly the important of bottom-up factors limiting bear populations, and especially the value of complementary diets.

This paper was published in the journal Oikos.

S. Nielsen, October 20, 2017

Citation:

Nielsen, S.E., Larsen, T.A., Stenhouse, G.B. & Coogan, S.C.P. (2017) Complementary food resources of carnivory and frugivory affect local abundance of an omnivorous carnivore. Oikos 126:369–380.

[ace-lab]

Spatial heterogeneity of the forest canopy scales with the heterogeneity of an understory shrub based on fractal analysis

Environmental heterogeneity characterizes the majority of natural systems, affecting animal-resource interactions among other aspects, but can be difficult to study given its scale-dependence. Identifying ways in which heterogeneity could be estimated based on broader landscape features readily assessed with remote sensing would be useful, especially for key food resources utilized by species at risk to inform conservation and management. For a chapter of her thesis, Catherine Denny (MSc. 2016) examined how spatial patterns in the forest canopy influenced those of Canada buffaloberry (Shepherdia canadensis) shrubs, which represent a crucial fruit resource for grizzly bears (Ursus arctos) in Alberta as they build body fat prior to winter.  Understanding their spatial pattern is therefore a high priority for understanding grizzly bear habitat.

To examine this question in more detail, Catherine used fractal analysis to evaluate overstory-understory relationships in buffaloberry across 20 km of transects (10 transects each 2 km long) where this shrub species was mapped at a fine scale (200,000 segments per 2-km transect). Buffaloberry patterns were indeed following fractal patterns with buffaloberry heterogeneity, defined by patch distribution and abundance, significantly related to evergreen canopy heterogeneity but unrelated to that of deciduous canopy. Effects of canopy on buffaloberry shrub presence also varied in strength with scale, and depended on canopy composition (deciduous vs. evergreen). Her findings highlight the importance of spatial scale and canopy composition in understanding the dynamics of shrubs in forests and in this case habitat (food supply) for wildlife like grizzly bears. This work also suggests that understory patterns may be predicted from forest canopy data obtained at larger landscape scales.

The Denny & Nielsen (2017) publication that describes this work can be found here.

C. Denny & S. Nielsen (October 2017)

Publication citation:

Denny, C.K. & Nielsen, S.E. 2017. Spatial heterogeneity of the forest canopy scales with the heterogeneity of an understory shrub based on fractal analysis. Forests 8(5):146. doi: 10.3390/f8050146

[ace-lab]

Boreal ground-beetle (Coleoptera: Carabidae) assemblages of the mainland and islands in Lac la Ronge, Saskatchewan, Canada

Aaron Bell published his first chapter from his MSc thesis on ground-beetle assemblages on Lac la Ronge in the Canadian Entomologist. Specifically, he examined how beetle communities on islands differed from those on the mainland and it turns out that only the smallest islands had communities that differed from the mainland. Surprisingly (or unsurprisingly perhaps if you're a fan of niche theory), these patterns were best explained when the characteristics (traits) of the beetles themselves were considered and more specifically the two traits of body size and mobility (winged or non-winged species). If you're large-bodied and can't fly, the smallest islands are a tough-go; perhaps because the habitat is less suitable and the only means to colonize islands is by drifting through the water. On the other hand, if you're small-bodied and can fly, the smallest islands are where it's at; maybe also because there's fewer big beetles around that might eat you.

Aaron Bell If you're interested in reading our article, you can access it here.

Bell, A.J., Phillips, I.D., Nielsen, S.E., Spence, J.R. 2017. The Canadian Entomologist.  DOI: 10.4039/tce.2017.12

Aaron Bell, Iain Phillips, Scott Nielsen, John Spence — at Lac la Ronge.

[ace-lab]

Where should we prioritize biodiversity conservation under climate change?

As most regions of the earth transition to altered climatic conditions, new methods are needed to identify the most likely refuges for biodiversity and to prioritize conservation actions. A variety of metrics and approaches have been proposed. Some are based on predicting future climates and rates of change (“climate velocity”). Others use only information on the current environment, finding areas where there are steep elevation gradients or topographic variation (“environmental diversity”) that help species to find climate refuges nearby. Faced with high stakes and a wide array of conservation targets, planners and land managers need new tools to deal with these new challenges.

In a new open-access paper published in Global Change Biology, led by Carlos Carroll and co-authored by U of A researchers Diana Stralberg, Scott Nielsen, and Andreas Hamann as part of the AdaptWest initiative, we set out to compare a variety of velocity and diversity metrics for conservation planning under climate change across North America. Specifically, we evaluated similarities and differences among different methods across different spatial scales and elevation ranges. Not surprisingly, we found substantial variation among metrics. But somewhat remarkably given uncertainty around future climate change projections, there was more variation among environmental diversity metrics based on current environmental conditions than among climate velocity metrics based on alternative future climates. We also found that while all diversity and velocity metrics generally increase with elevation, so do the contrasts among them, due to interactions between climate and terrain (see figure below).

So what is a planner to do, given all these differences? We suggest that metrics be combined, with areas of greater variation down-weighted (all spatial data are being made available through AdaptWest). Alternatively, finer-scale diversity metrics can be substituted where available, and supplemented with data on key target species as needed. Climate velocity metrics are useful for identifying broad-scale “macro-refugia,” where more species may find a long-term refuge from climate change. Areas of high environmental diversity should correspond with greater potential for local “micro-refugia” that can serve as temporary havens for species under a climate in flux. Where they coincide, short- and long-term conservation potential can be achieved most efficiently. We found that neither type was well-represented by the current protected area system, suggesting that much conservation work is still needed in order to prepare and adapt where possible to climate change.

D. Stralberg

Citation: Carroll, C., Roberts, D.R., Michalak, J.L., Lawler, J.J., Nielsen, S.E., Stralberg, D., Hamann, A., McRae, B.H., Wang, T. 2017. Scale-dependent complementarity of climatic velocity and environmental diversity for identifying priority areas for conservation under climate change. Global Change Biology (early view).

Link to paper: http://onlinelibrary.wiley.com/wol1/doi/10.1111/gcb.13679/abstract

[ace-lab]

Lakes fragment and diversify fire regimes

There is a commonly used saying that ‘the boreal forest is born of fire’. Species here are generally well-adapted to dealing with periodic, stand-replacing fires typical of the biome with a pool of species evolved to deal with fire in either time or space. For instance, most boreal plants reproduce through vegetative re-sprouting or in the case of jack pine and black spruce through aerial seedbanks housed within serotinous cones that open after fire. This makes these systems quite resilient, if not dependent, on fire.

Species that are more sensitive to fire can take advantage of the sheer scale of the boreal biome. At broad spatial scales and over long enough periods the entire system of apparently stochastic disturbances approaches a quasi-equilibrium state where everything basically averages out, but there is still great heterogeneity in local stand ages. New sites burn, others age to old growth. In recent years, however, the frequency of fires has steadily increased despite cyclical patterns in fire being the norm. In fact, in some places, like in Alaska’s boreal forest, fires have doubled in the last 50 years compared to the prior 3,000 years. 

Likewise, Canada’s boreal forest appears to be in the mist of similar increases in fire frequency and possibly fire size, particularly within the western boreal forest. Very large fires, like that of the Horse River fire of Alberta in the spring of 2016 or the Richardson fire just north of the Horse River fire only 5-years prior, provide two examples of how extreme and large these fires can get (both fires combined are larger than the state of Connecticut).

Although fires are natural to the system and region, in an increasingly warming world associated with increases in greenhouse gas (GHS) emissions the area burned in Canada is predicted to increase further. For instance, the two fires mentioned above totaled nearly 30% of Canada’s annual emissions or 4-times the amount produced per year by oil sands (assuming average forest fire emission rates). The boreal forest’s massive carbon stores (sinks) are increasingly become a source of GHG emissions further contributing to warming and thus more fires.

One direct outcome of all of this is less old-growth forests and thus less of a quasi-equilibrium condition over the short term. In other words, it may be a bad time to be an old-growth forest-specialist in the western boreal, even in areas free of human activity, such as the Canadian Shield in northern Saskatchewan.

Although we normally think in conservation of forest fragmentation as being negative, natural patterns in non-forest habitat, such as lakes within the boreal forest, may actually maintain old-growth specialists by locally reducing fire return intervals. In particular, large lakes can impose fire breaks that result in unburned fire skips providing refuge for old growth conditions. Although this fire ‘shadow’ or fire ‘refugia’ concept is well-known conceptually and often discussed qualitatively, less is known on how specifically landscape patterns in lakes (and which measures) interact to promote fire shadows/refugia and thus forest age patterns.

Our new paper in Forests addresses this question by examining how patterns in large lakes in Saskatchewan’s boreal forest act as firebreaks. Although there is still much stochasticity in where fires burn, distinct fire shadows are evident around the periphery of large lakes that depend on not only lake size and distance from lakes, but also the shape of large lakes, direction from large lakes, and amount of surrounding water (other lakes). These factors interact in complex ways promoting a greater diversity of possible fire regimes and thus stand ages that depend on landscape structure of lake patterns and thus natural fragmentation of forest fuels. More specifically, we found that lake sizes greater than 5,000 ha were particularly effective at reducing the likelihood of a site burning near lakeshores. Depending on other landscape conditions, fire rotation periods around these lakes were predicted to vary 15-fold, from as frequent as 47 years when away from lakes to more than 700 years when adjacent to large, complex lakes and to even longer fire rotation periods on lake islands.

These results point to a strong, but complex, bottom-up control of local wildfire activity that is based partly on the configuration of lakes acting as firebreaks. Thus, the chance of an individual site burning is not just due to weather and chance alone that dominates the current viewpoint of many, but also landscape patterns of the forest and in this case lakes. The fragmentation of forests from lakes therefore not only makes for a more beautiful landscape of water and land, but also a more diverse and heterogeneous forest on the land that is more likely to contain old growth forest characteristics along its lake edges.

SE Nielsen

Citation: Nielsen, S.E.; DeLancey, E.R.; Reinhardt, K.; Parisien, M.-A. 2016. Effects of lakes on wildfire activity in the boreal forests of Saskatchewan, Canada. Forests 7(11):265.

Link to paper: http://www.mdpi.com/1999-4907/7/11/265

[ace-lab]

Grizzlies Face Ecological Trap

 Clayton Lamb (PhD Candidate, ACE Lab member) and colleagues recently published an article detailing an ecological trap for grizzly bears in southeast British Columbia.  An ecological trap is a habitat in which animals would have historically survived or reproduced well in, but no longer do because of rapid human changes on the landscape.

Clayton and colleagues used multiple sources of data for their investigation, including DNA fingerprints from 500 grizzly bears collected over 8 years of non-invasive hair sampling and 1800 vegetation plots.

The key results of the study show that grizzly bears in the ecological trap area had abundant fruit resources (huckleberry and buffaloberry, as well as other foods such as salmon, road killed ungulates and human foods), yet bears occupying this area faced 17% lower survival rates compared to adjacent areas. The effects of this attractive habitat misleading the bears habitat selection was not only low survival for individuals, it also contributed to broad-scale population declines because bears from secure, backcountry areas were moving into the ecological trap as individuals were killed and new space was opened.

The article was published by the Journal of Animal Ecology and can be found here:

http://onlinelibrary.wiley.com/doi/10.1111/1365-2656.12589/full

 What this means is that human habitation in productive grizzly bear habitat has consequences for both humans and bears due to intense human-bear conflicts. For humans, living in grizzly bear country necessitates increased effort to reduce conflicts including managing attractants and increasing non-lethal management.  For bears, the localized mortality that occurs where humans and productive habitat overlap have far-reaching demographic consequences in the form of attracting dispersing individuals from backcountry regions. Within our study area we have document displacements of individuals, who were killed in the ecological trap, but had at one time lived nearly 60 km away in the backcountry. Even more broadly, collared animals from both Alberta and the United States have recently showed up in the ecological trap, highlighting the spatial scale that these processes can act on.

 Animal populations are becoming increasingly fragmented and foundational work by Dr. Michael Proctor has shown that grizzly bear populations are fragmented by human settlement and highways across North America. Our work builds on this idea and documents the causal mechanism severing population connectivity of grizzly bears in the Southern Rocky Mountains.

  This work was cited in the recent re-assessment of brown bears across the globe (see “download assessment”):

http://www.iucnredlist.org/details/41688/0

This work has attracted much interest since first appearing online in September 2016, including a fantastic infographic from @MammalLady on Twitter:

https://twitter.com/search?q=%40mammalLady%20trap&src=typd

 This work has already been featured in University classrooms across western North America:

Dr. Sophie Gilbert, U of Idaho https://twitter.com/SophieLGilbert/status/827613092427161600

Dr. Brian Starzomski, U of Victoria    https://twitter.com/BStarzomski/status/791732665678385152

[ace-lab]

Fen-tastic views from the air over McClelland Fen in Northern Alberta!!