Thursday, July 23, 2015

Plight of the Bumble Bees: How shabby climate analyses and lax peer review promote a dreadful remedy


 In July 2015 the journal Science published Kerr et al’s Climate Change Impacts On Bumblebees Converge Across Continents. It was a woeful analysis hyped by the media. It did very little to further our understanding of the causes of bumblebee declines and more likely obscured the real problems. But it did illustrate why the public is becoming increasingly suspicious of “scientific claims” regards catastrophic climate change as well as demonstrating the inadequacy of the peer review process.

There were 4 major problems.

1) By employing a skewed statistical methodology and using inappropriate metrics, Kerr 2015 contradicted the biologists’ consensus (Goulson 2015) to argue bumblebees declines are independent of land use changes, pesticides and introduced pathogens.

2) Kerr 2015 results demonstrated that bees are not tracking their climate niche and are not responding to climate changes as predicted. Their data strongly suggests range shifts have been independent of climate change. Either the bees are insensitive to decades of climate change or climate change has had little impact on the bees’ critical microclimates. Nonetheless based on bad statistical modeling, they claimed range shifts were “independent of changing land uses or pesticides”, and then spun a climate catastrophe scenario by simply asserting the default cause must be climate change.

3) Kerr 2015 totally ignored the leading hypothesis that points to introduced pathogens as the cause of sudden declines and shifts in a select group of related North American Bees (Cameron 2014, The Xerces Society 2008). Kerr’s climate interpretation suggests transporting bumblebees to new northerly habitat, knowing it poses greater risks by spreading pathogens and further endangering susceptible species.

4) Kerr 2015 demonstrate that the journal Science strayed from objectivity into climate change advocacy. Science not only failed to properly edit this paper, they added an additional “news” commentary Bumblebees Aren’t Keeping Up With A Warming Planet and quote Kerr’s catastrophic view, “Climate change is crushing species in a vise”. The other global warming advocacy journal Nature ran a simultaneous apocalyptic story “Climate Change Crushes Bee Populationsencouraging wide spread media fear mongering.

1. Kerr 2015’s Inappropriate Statistical Methodology

The first statistical violation was Kerr’s categorization of time periods that prevented their models from accurately detecting the effects of land use change. They analyzed changes in bees’ latitudinal and thermal limits using records for 31 North American and 36 European species. To create a “pre-climate change” baseline for each species, they averaged 5, 10 or 20 extreme observations (depending on availability) for the time period 1901-1974. For example to determine a species’ most southerly latitude, they averaged the 5 most southerly records across the continent. However those averages would be dominated by the earliest decades and could hide any northward retractions that happened in the baseline’s later decades. To determine the bees’ warmest thermal limits, they likewise averaged 5 modeled temperatures from the warmest occupied sites. They similarly averaged observations restricted to 3 later 11-year periods of purported human caused climate change spanning 1975-1986, 1987-1998, and 1999-2010, and then compared those averaged results with the baseline averages.

However their asymmetrical categorization of a 74‑year baseline period vs. three 11‑year “climate change” periods is highly problematic. If their intent was to determine the timing of any significant shifts, their analysis should have compared equal decade-long periods.  Instead because their technique averaged the most extreme southern latitudes, the baseline would easily be dominated by the earliest 20th century observations. Any range retractions that happened later during the baseline period would not be “statistically detected” until the 1975-1986 “climate change” period. Any editor or peer reviewer should have required a correction, knowing their asymmetrical categorization could cause such misleading results.

Many researchers from both North America and Europe (Fitzpatrick 2007) have documented that the period between 1940-1960 encompassed the greatest shift in agricultural expansion and intensity that has gravely affected bee populations. For example, studies in Illinois (Grixti 2009) determined that the greatest loss of bumblebee abundance, species richness and shifting ranges occurred between 1940-1960 due to agricultural intensification. After 1960, only minimal shifts occurred for the following 2 decades as agricultural expansion waned. But Kerr’s baseline categorization would not detect those range shifts until the 1975-86 period. The resulting statistical illusion of their model then created the incorrect perception that major range shifts were independent of those agricultural changes. 

Kerr’s main paper only provided graphs for the final 1999-2010 period, so in my Figure 1 below, I have also added the 1975-1986 graphs from their supplemental data to also compare the recent decadal shifts. Oddly their results contradict their assertion that landscape restrictions were preventing bees from migrating, and therefore climate change was “crushing bees in a vise”. Their data clearly show half of the European species (green dots) were moving northward while most of the North American species (red dots) were shifting southward. In a NY Times’ interview, bumblebee expert Dr. Sydney Cameron also noted this lack of correspondence between assertions and evidence, diplomatically stating Kerr’s suggestion of thwarted northward migration was “a surprising conclusion given the data.” Clearly the bees are not caught in any such vise. The average shift in latitudinal positions was simply contradicting global warming theory.

Second if the 1940-1960s land use changes were the major driving factor, instead of climate change, we would expect dramatic range shifts in the 1975-1986 period, but only minor range shifts between 1975 and 2010. In contrast if climate change was the driver, we would expect increasing range shifts between 1975 and 2010 as purported climate change intensified. The data does not support a climate change interpretation.

The 2 graphs on Figure 1’s left (A’s) illustrate shifts in each species’ average extreme northern latitude, while the 2 graphs on the right (C’s), illustrate the change in their southern extremes. The X-axis represents the species latitudinal extremes in terms of distance (kilometers) from the equator during the base-line period. (Figure 2 helps the reader visualize the geographic location for those distances.) The Y-axis represents the species latitudinal deviation from the baseline period. A positive number means the species’ extreme latitude shifted northward and a negative number means it shifted southward. The dashed line at “0” represents the 1901-1974 base line latitude. Species that have not shifted their latitudinal margins will be located on that dashed lines.

For example, I added blue arrows to highlight that one European species’ northern-most latitude, originally located about 6400 km north of the equator (X-axis), had already shifted northwards by 1000 km (Y-axis) by the 1975‑period. Assuming the second arrow points to the same species, there was no further shift through the 1999-2010 period, suggesting no effect from recent climate change. Readers should also note that a majority of the species on both continents had retracted their northern limit southwards by the 1975-1986 period, again the opposite of what global warming predicts. By 1999-2010, half the species still exhibited ranges that had retracted southwards, although there was a slight increase in species that expanded northward.

Bumblebee northern and southern extreme range changes

The graphs on Figure 1’s right side represent shifts in the species most southerly margins. Again the bees are shifting differently on each continent, suggesting regional drivers, not global climate change. Because Kerr’s graphs have a different scale, I added a blue line to highlight any northerly retraction exceeding 400 km.  By the 1975-period (top right), nearly all the North American species (red dots) had already retracted their southern range northward to some degree. By the 1999-2010 period, the greatest North American retraction remained at 1000 km, while 3 species expanded their range southward, again contradicting a global warming interpretation. The remaining North American latitudinal shifts are not noticeably different between 1975 and 2010. Furthermore, it should be noted that any retractions in the southeast USA are probably not linked to global warming because most of that area has been deemed a “warming hole” with a 20th century cooling trend for maximum temperatures (see Fig 13 Menne 2009).

In Europe (green dots), half the species had expanded southward by the 1975-1986 period again contradicting global warming theory. By the 1999-2010 period more species began retracting northwards while the 2 most northerly species move southward retracing their earlier retractions. Because some declining species have shifted northwards while others shifted southward, most European researchers had rejected the hypothesis that climate change has been driving declining bee populations. (Willliams 2007)

Unfortunately from Kerr’s results, we cannot determine which dot represents which species, and thus we are prevented from using additional research that might elucidate why an individual species shifted its range when another species did not. Meta-analyses such as this only create average trends from a lumped set of species but typically obscure the variety of confounding factors that may be driving these diverse and complex range shifts. Yet such meta-analyses are often the preferred method for researchers advocating climate change disruption because they assume the variety of confounding factors cancel out, leaving only a climate change footprint (Dr. Singer, personal communication) A problematic IPCC meta-analysis is discussed here.

In addition to skewed temporal categories, Kerr 2015 used an inappropriate metric to dismiss land use changes. Kerr compared recent satellite data with past characterizations of the landscapes to determine changes in cropland and pasture extent. But extent, or acreage, is not the only land use factor that could impact bees. The major factor is the loss of flowers.

Due to cheaper synthetic fertilizers, many croplands no longer plant crops of bee-nourishing alfalfa to rotate with crops of wind-pollinated corn or wheat. Planting alfalfa had partially offset the loss of flowers when native grasslands were cultivated. Additionally pastures and grasslands are managed to reduce insect pollinated flowers and promote more wind‑pollinated grasses.

Furthermore methods for producing silage have increasingly replaced traditional hay‑making. Traditional hay‑making requires a good stretch of dry weather that lowers the hay’s water content, so mowing typically occurs in late summer. In contrast silage fermentation requires greater water content than hay, so fields are mowed earlier and sometimes more often. Earlier mowing removes nourishing flowers so bee species that emerge later in the season from “hibernation” are critically impacted (Fitzpatrick 2007). Additionally wind-pollinated corn has increasingly become a major source of silage replacing alfalfa and soybean.

These agricultural practices have increased production over the past few decades without cultivating more land, so those land use changes would not be detected as changes in cropland or pasture “extent”. But those changes most certainly impact bees.  Again any editor or peer-reviewer familiar with the plight of the bumblebees should have been aware that “extent” was likely a meaningless metric. Yet by using the “extent” metric, Kerr’s models incorrectly asserted that landscape changes had no impact, contradicting a wealth of research demonstrating a heavy toll by landscape changes.

Latitude in terms of kilometers north of equator

Still Kerr schizophrenically embraced landscape changes to help explain why so many bee species had contradicted climate change theory by shifting to lower elevations (Figure 3 below). Bees that moved to higher elevations were touted as confirmation of climate change induced shifts. But to dismiss the contradictory evidence, Kerr 2015 nebulously suggested global warming could increase forest growth at higher elevations and that resulting landscape change could eliminate bee habitat thus forcing bees to lower levations. But that begs the question of why half the bees still migrated to higher elevations. Reforestation may eliminate some warm sunny bee habitat, but in Europe the dominant cause of reforestation has been the abandonment of marginal farmlands (Gehrig-Fasel 2007). Furthermore the downward shift in elevation seen in Europe’s high latitude bee species is consistent with Scandinavian tree ring data that suggests temperatures have been cooler since the 1950s (Esper 2012). In agreement with “cooling” tree rings, many butterflies in Finland that had expanded northward during peak warming of during the 1930s to 50s, have also retreated southward. (Poyry 2009).

Bumblebee Elevational Changes

In North America, many bee species have also moved to lower elevations in the most recent decades (Figure 3) and this is consistent with shifts to lower elevations by several other species. In the United States vegetation in the Sierra Nevada has been moving down‑slope (Crimmins 2011). Montane butterfly populations that Parmesan claimed had gone extinct due to global warming have now returned and there is no longer a statistical shift to higher elevations (discussed here). A high percentage of newly discovered pika populations have been observed at much lower elevations than had been observed during the 1920s (discussed here). And mirroring bumblebees’ shifts, 20% of California’s bird species have moved upslope, while 20% moved down‑slope while most have not shifted at all during the 20th century (Tingsley 2012).

2. Bumble Bees Move Independently of Climate Change

Assuming that species are in equilibrium with their environment, ecologists infer a species’ temperature tolerances based on the most extreme temperatures throughout their range and then construct a bioclimatic envelope. However the usefulness of bioclimatic envelopes has been increasingly debated (Hampe 2004) and Kerr’s data demonstrates why. Theory predicts that if a habitat warms or cools, species must shift in order to remain within their temperature envelope’s boundaries. In Kerr’s graph below (Figure 4), the dashed line, at zero on the Y-axis, represents each species’ baseline limit for cold temperature tolerance (Fig. 4’s graphs on left, B’s), and for warmth tolerance (graphs on right, D’s).

Bumblebee Changes in Extreme Temperature LImits

If a species’ range tracked its thermal limits, its representative dot would sit on the dashed line. Any dot above that line means they have retreated to warmer habitat. Any dot below the dashed line means the species retreated to cooler habitat. The X-axis represents the species thermal limit determined by the base line period. For example, for species’ extreme cold limits, several species persisted in regions experiencing winter extremes of  -10°C during the base line period (X-axis).  But during all the later periods, the coldest temperatures experienced by most species were 2 to 6 degrees warmer, (-8 to -4°C). So the bees are said to be lagging climate change because they are remaining in warmer regions.

Regards the bees’ extreme warm limits, the opposite is happening for most species. Nearly all of North America’s species (red) retracted their range by 1975 and inhabit much cooler regions than required by their bioclimatic envelope. Those bees now inhabit regions where maximum temperatures are 1 to 12 degrees cooler than their baseline period. In contrast, many European species expanded into warmer regions although the majority also retracted to cooler areas. With few species sitting on the dashed line, the data clearly shows most bee species are not tracking climate change and have shifted their ranges independently of calculated thermal limits. An alternative interpretation would argue the baseline observations never accurately defined the bioclimatic envelope. Whatever the case, clearly factors other than climate were forcing bees to alter their thermal ranges.

3. Failure to Address Pathogen Spillover Hypothesis

In North America a few closely related species in the same subgenus began a rapid decline in the late 90s. Abundance declined by up to 96% and geographic ranges contracted by 23‑87%, mostly within the last 20 years (Cameron 2011). Species once designated as abundant or common, declined to being rare or absent in just 7 to 10 years. In addition to the rapid decline, only certain species were affected while others remained abundant. So many researchers rejected climate change as a causative factor and suggested the importation of a novel pathogen was the likely cause (Thorp 2008). Commercially grown bumblebees were being transported around the world, and in the late 90s North American native bees, were reared for commercial purposes in European facilities and then re-introduced to America. Those species are believed to have been infected by a novel pathogen that they introduced to North America. One species, Bombus occidentalis that widely inhabited western North America, began a sudden sharp decline at the same time commercially raised B. occidentalis populations in greenhouses were also exhibiting declines due to the parasite Nosema bombi. Shortly thereafter two other closely related bee species began to rapidly decline. By 2010, over 60 top bee biologists petitioned the USDA’s Animal and Plant Health Inspection Service to regulate the commercial bumblebee industry to ensure transported bees were disease free.

Although the specific strain of pathogen driving these observed declines has not been determined with full certainty, there has been growing support for the pathogen hypothesis as declining species are observed to harbor heavier pathogen loads than stable bee populations (Cameron 2011, Szabo 2012, Colla 2006, Malfi 2014).

Understanding and preventing the spread of deadly disease should be a major societal focus because it severely affects all species. Introduced pathogens wreaked havoc in the Americas ever since Europeans brought smallpox to the western hemisphere and decimated Native American populations. More recently, an introduced chytrid fungus has inflicted a wave of global amphibian extinctions. An introduced European fungus is now  decimating eastern USA bats. In the 80s, scientists were transporting the African Clawed Frog around the world to use in pregnancy testing and embryological studies. The African Clawed Frog harbors the deadly chytrid but is unaffected by it and so served as a carrier. As the fungus was inadvertently spread to new environments, susceptible species like Costa Rica’s Golden Toad and other closely related species rapidly went extinct. While ecologists embarked on efforts to minimize the spread of the disease and save the most vulnerable amphibian species, one of the IPCC’s specially selected biologists, Alan Pounds, denigrated those efforts because he falsely believed the extinctions were a result of catastrophic climate change (discussed here). He oddly argued that by blaming the pathogen, scientists were redirecting the public’s attention from addressing a speculative CO2 climate catastrophe. But Pounds’ remedy, reducing our carbon footprint, would never have stopped the spreading disease, nor saved a single frog and CO2 advocates were hindering the development of real solutions. Likewise controlling our carbon footprint will do precious little to remedy the plight of the bumblebees.

Not only does Kerr 2015 completely ignore the devastating impacts of introduced pathogens, their climate change remedy argues for transporting species northward into habitats where global warming models suggest species should have shifted. In contrast, in a NY Times interview, bumblebee expert Dr. Sydney Cameron took issue with Kerr’s suggestion that we should intervene with “assisted migration”, because that remedy risks spreading pathogens.

Dr. James Strange added. “I did not come away convinced that climate change is causing these movements.” Strange also worries that Kerr 2015 might cause people to blame climate change entirely for bee population destruction and ignore potential factors such as parasites, pesticides and habitat destruction. “There’s a bit of me that’s nervous someone will pick this up and say ‘They figured it out: It’s climate change,’ ” Dr. Strange said. “But really, we haven’t figured it out yet.”

Indeed Dr. Strange should be concerned. If there is anything we have learned from the Golden Toad extinctions, Edith’s Checkerspot extirpations, or the Emperor Penguins, advocacy for CO2 caused catastrophic climate change has blinded people from all walks of life to the more urgent conservation issues.