Research Article |
Corresponding author: H. E. James Hammond ( james.hammond@canada.ca ) Academic editor: Thorsten Assmann
© 2021 H. E. James Hammond, Sergio García-Tejero, Greg R. Pohl, David W. Langor, John R. Spence.
This is an open access article distributed under the terms of the CC0 Public Domain Dedication.
Citation:
Hammond HEJ, García-Tejero S, Pohl GR, Langor DW, Spence JR (2021) Spatial and temporal variation of epigaeic beetle assemblages (Coleoptera, Carabidae, Staphylinidae) in aspen-dominated mixedwood forests across north-central Alberta. In: Spence J, Casale A, Assmann T, Liebherr JК, Penev L (Eds) Systematic Zoology and Biodiversity Science: A tribute to Terry Erwin (1940-2020). ZooKeys 1044: 951-991. https://doi.org/10.3897/zookeys.1044.65776
|
Epigaeic beetle assemblages were surveyed using continuous pitfall trapping during the summers of 1992 and 1993 in six widely geographically distributed locations in Alberta’s aspen-mixedwood forests prior to initial forest harvest. Species composition and turnover (β-diversity) were evaluated on several spatial scales ranging from Natural Regions (distance between samples 120–420 km) to pitfall traps (40–60 m). A total of 19,885 ground beetles (Carabidae) representing 40 species and 12,669 rove beetles (non-Aleocharinae Staphylinidae) representing 78 species was collected. Beetle catch, species richness, and diversity differed significantly among the six locations, as did the identity of dominant species. Beetle species composition differed significantly between the Boreal Forest and Foothills Natural Regions for both taxa. Staphylinidae β-diversity differed significantly between Natural Regions, whereas Carabidae β-diversity differed among locations. Climate variables such as number of frost-free days, dry periods, and mean summer temperatures were identified as significant factors influencing beetle assemblages at coarse spatial scales, whereas over- and understory vegetation cover, litter depth, shade, slope, and stand age influenced beetle assemblages at finer spatial scales. Significant interannual variation in assemblage structure was noted for both taxa. Because composition of epigaeic beetle assemblages differed across spatial scales, forest management strategies based only on generalized understanding of a single location will be ineffective as conservation measures. In addition, site history and geographic variation significantly affect species distributions of these two beetle families across the landscape. Thus, we underscore Terry Erwin’s suggestion that biodiversity assessments focused on species assemblages at different spatial scales provide a sound approach for understanding biodiversity change and enhancing conservation of arthropod biodiversity.
Climate, forest insect assemblages, insect biodiversity, spatial scale, variance partitioning, vegetation
Terry Erwin’s studies of the flora and fauna of tropical forests have profoundly influenced global efforts to understand forest biodiversity. In his classic paper that has inspired people to pursue questions about insect biodiversity
Unlike the tropical forests that Terry studied, where overstory tree richness often exceeds 200 species per hectare (
The typical upland mixedwood succession pattern in western Canada is for deciduous species, mainly trembling aspen, to dominate upland stands as they recover from disturbance. Such aspen-dominated forests eventually give way to variable mixes of deciduous and coniferous trees and finally to conifer-dominated (usually white spruce) stands in the prolonged absence of wildfire and depending on availability of conifer seed sources (
Boreal mixedwood stands provide extensive habitats for a wide range of forest biodiversity in western Canada. Although taxonomy, distribution, and habitat affinity of some groups such as plants and vertebrates are well understood, relatively little information is available for hyper-diverse groups such as invertebrates. In fact, only the small subset of species that cause economic damage to trees, biting insects, and a few charismatic groups such as butterflies and dragonflies are reasonably well known. Thus, the poor state of knowledge about most of the biodiversity in these forests is especially concerning as these landscapes are widely impacted by cumulative effects of timber harvest, oil and gas exploration, climate change, and changing fire cycles and intensity (
Amongst terrestrial invertebrates, epigaeic beetles, especially Carabidae, have been popular subjects for disturbance ecology research globally (
One of the earliest studies, conducted in 1992–1993, addressed spatial variation in the structure of epigaeic beetle assemblages in aspen-dominated forests of Alberta, as extensive commercial harvesting of aspen had already begun. The most urgent conservation-oriented aspect of this work focused on whether old growth aspen-dominated forests (> 100 years post-fire) harbored significantly different assemblages than did mature forests (50–80 years) (
In this paper, we examine composition and structure of carabid and staphylinid assemblages to determine (1) whether Natural Regions, Subregions, and stand age are useful predictors of arthropod assemblage structure, (2) β-diversity within Natural Regions, Subregions or stands of the same age at different spatial scales, (3) the main patterns of arthropod community composition at different spatial scales and the extent to which they are explained by environmental variables, and (4) how such information can best contribute to effective arthropod conservation.
Assemblages of Carabidae and Staphylinidae were sampled from the forest floor in sites from six locations in Alberta (Fig.
Summary of forest stand characteristics of aspen mixedwood forests sampled by pitfall trap lines at six locations in north-central Alberta, 1992–1993.
Natural Region | Natural subregion | Location | Stands | General location | Stand age | Stand | Stand |
---|---|---|---|---|---|---|---|
Latitude / Longitude | (years) | size (ha) | elevation (m) | ||||
Boreal Forest | Dry Mixedwood | George Lake | GLED | 53.9564°N, -114.1233°W | 80 | 60 | 693 |
Boreal Forest | Dry Mixedwood | George Lake | GLMC | 53.9575°N, -114.1194°W | 80 | 130 | 687 |
Foothills | Lower Foothills | Hinton | HIA | 53.33308°N, -117.5977°W | 80 | 5 | 1151 |
Foothills | Lower Foothills | Hinton | HIB | 53.4864°N, -117.3825°W | 75 | 100 | 1059 |
Foothills | Lower Foothills | Hinton | HIC | 53.50452°N, -117.32148°W | 85 | 20 | 1162 |
Boreal Forest | Central Mixedwood | Lac la Biche | M2 (2 lines) | 54.8421°N, -111.4919°W | 52 | 269 | 667 |
Boreal Forest | Central Mixedwood | Lac la Biche | M3 (4 lines) | 54.8349°N, -111.657°W | 51 | 315 | 642 |
Boreal Forest | Central Mixedwood | Lac la Biche | O2 (4 lines) | 54.8458°N, -111.5873°W | 125 | 134 | 649 |
Boreal Forest | Central Mixedwood | Lac la Biche | O4 (2 lines) | 54.8531°N, -111.4473°W | 122 | 187 | 691 |
Boreal Forest | Lower Boreal Highlands | Peace River | PRA | 56.4131°N, -117.7308°W | 80 | 38 | 731 |
Boreal Forest | Lower Boreal Highlands | Peace River | PRB | 56.4042°N, -117.6856°W | 70 | 37 | 710 |
Boreal Forest | Central Mixedwood | Rose Creek | RCA | 53.05°N, -115.0578°W | 105 | 10 | 819 |
Boreal Forest | Central Mixedwood | Rose Creek | RCB | 53.0433°N, -115.1361°W | 100 | 25 | 860 |
Boreal Forest | Central Mixedwood | Rose Creek | RCC | 53.0428°N, -115.0714°W | 95 | 10 | 835 |
Boreal Forest | Central Mixedwood | Slave Lake | SLA | 55.2411°N, -114.7036°W | 70 | 25 | 604 |
Boreal Forest | Central Mixedwood | Slave Lake | SLB | 55.3372°N, -114.9567°W | 125 | 20 | 603 |
We sampled mesic forest stands that were primarily dominated by Populus species, but which also contained mixes of scattered white spruce, paper birch, lodgepole pine, and willow (Suppl. material
Epigaeic Carabidae and Staphylinidae were sampled using lines of six pitfall traps (
Most stands were sampled with one linear transect or ‘line’ of six traps. Within a line, traps were separated by ~ 50 m, which is sufficient to ensure independence of catch (
All adult carabids and staphylinids were identified to species using taxonomic literature (primarily Lindroth 1961–69 for carabids; primarily
During the summer of 1992 we measured several site variables at trapping sites in each stand in order to test whether such features were correlated with beetle catches. Overstory and understory plant communities were assessed around each pitfall trap using a 5 m × 5 m vegetation plot centered on the trap. Tree cover was estimated as the number of stems > 5 m in height whose driplines fell into the 5 m × 5 m vegetation plot. Shrub density was measured by counting the shrub stems occurring at 0.5 m off the ground, in two 1 m wide transects, 5 m in length, one running north to south and the other running east to west, with the trap at the midpoints of both transects. Litter biomass (dried) was determined from a representative 0.25 m2 area chosen with the vegetation plot. We calculated other site characteristics such as southern aspect and slope using digital elevation models (DEM) downloaded from the Federal Geospatial Platform of the Government of Canada (url: https://ftp.maps.canada.ca/pub/nrcan_rncan/vector/index/html/geospatial_product_index_en.html#link). We calculated the southern aspect of the terrain around each trap as a proxy for solar irradiation, measuring how many degrees the orientation deviated from north, going from 0 (north) to 180 (south). Two different variables were computed: one considering only slopes > 2° and the other > 5°; terrain with slope < 2° or < 5° were considered as facing south and given a value of 180. Climatic data were downloaded from http://albertaclimaterecords.com/, which summarizes 30-year climatic averages from 1981 to 2010 on a 10 ×10 km provincial grid. These variables are summarized in Suppl. material
Our entire dataset contains a total of 24 pitfall trap lines; however, because we sought to study several levels of variation in beetle assemblages, we separated the data into two subsets. The regional dataset contains all lines at all locations, except for Lac la Biche, where instead only four of the lines (two mature and two old from the 12 lines, Fig.
Sampling effort varied among locations (range 62–176 days), due to different setup dates and losses attributable to trap disturbance by wildlife. Thus, the catch of each beetle species was standardized to 175 trap days [(raw abundance * (175 / actual number of trap days)]). These standardized abundance data were then subjected to the ‘Hellinger’ transformation, which reduces the ‘double zero problem’ of resemblance among samples, before preparing the data for linear analyses based on Euclidean distances (
Differences in standardized abundance between years, ecoregions and locations were examined using a generalized linear mixed model (function glmer in R). Each trap catch was classified into a year, ecoregion and location, and formal hypothesis testing was done using pitfall trap line as a random effect. Effects were tested using Poisson, quasi-Poisson, and negative binomial distributions, and for each taxon the results were congruent for all error distributions based on Analysis of Deviance using Type II Wald Chi-square tests. In this paper we report the results based on the negative binomial model. Multiple comparisons tests were conducted using Tukey’s contrasts with a Holm adjustment to control experiment-wide error rate.
Beetle species richness among locations was compared using coverage-based rarefaction (
Spatial and temporal patterns of assemblages were examined using Principal Components Analysis (PCA). PCA reduces the dimensionality of multivariate data by constructing principal components, which are new variables that are uncorrelated, linear combinations of the original data. Principal components explain the maximum amount of variance in our community data in the fewest dimensions. In addition, interannual variation in the carabid and staphylinid assemblages was examined using pairwise Bray-Curtis measures of dissimilarity, transformed to percent similarity (1 – dissimilarity), on the standardized catch of the entire fauna, using species with a catch of greater than five individuals, and to presence-absence data for all species.
Redundancy Analysis (RDA) was used to ascertain if Natural Regions and Subregions (regional dataset) and stand age (local dataset) can be useful tools to classify the carabid and the staphylinid assemblages. The results were tested using 4999 permutations of the residuals, which were restricted to account for the nested model and fine-scale spatial autocorrelation (i.e., traps within lines were permuted along series in order to keep their spatial arrangement). When these classifications proved useful, we looked for significant indicator species for each group following the approach in
Multivariate dispersion within each Region, Subregion, location, and line, and within each stand and each line for mature and old stands at Lac la Biche, was used to examine within-group β-diversity. We then tested for differences between Natural Regions, Subregions or stands of different age using restricted permutations (see above).
To explore the main patterns in the data we used multivariate regression trees (
Variables about climate and habitat structure were used to explain patterns at the Natural Region scale. We first forward selected the climatic variables in RDA according to the amount of variation that each could explain and tested the results using restricted permutations (see above). Only the first two variables were kept in the model to avoid overfitting, and since our sites are clustered, there were only eleven different values for each climatic variable. Keeping these two variables in the model, we included habitat structure variables by forward-selection. Tests in this second case were done without restricting the permutations since environmental variables were measured at each trap and we expect that they may be the cause of possible spatial autocorrelation patterns in the beetle data (i.e., we assume the “environmental control” model, see
Variation partitioning was used to combine results from environmental RDA models, together with dummy variables coding for locations and lines in the regional dataset, and for stands and lines, in the Lac la Biche dataset. This allowed us to better understand how the main patterns of variation are arranged across spatial scales and the extent to which they can be explained by environmental variables.
All beetle species and community analysis were computed with RStudio running R version 4.0.2 (
A total of 32,554 epigaeic beetles was collected during the summers of 1992 and 1993, of which 61% were Carabidae and 39% Staphylinidae (excluding Aleocharinae; Suppl. material
Boxplot of the standardized catch of epigaeic beetles at seven locations in aspen-dominated mixedwood forests in north-central Alberta, 1992–93. Locations are distributed across four Natural Subregions (SR) and two Natural Regions (NR) A Carabidae B Staphylinidae. Locations with the same letter situated above the bar are not significantly different at α = 0.05 (post hoc Tukey’s tests). For locations where there was significant interannual variation, the results are indicated above the bar for that location.
Of the 40 species of Carabidae and 78 species of Staphylinidae captured across all locations and years, all but three species are native. The non-native Palearctic carabid, Pterostichus melanarius (Illiger, 1798), was collected at two locations in this study: George Lake (both stands; ca. 5% of total catch) and the HIB stand at Hinton (one specimen). The George Lake stands are embedded in an agricultural setting and P. melanarius was first documented as an invader of these stands in 1981 (
Rarefaction-estimated species richness of Carabidae was highest at Peace River in the Lower Boreal Highlands Subregion, despite the fact that catch was lowest there (Fig.
Coverage based rarefaction estimates of species richness and diversity of epigaeic beetles at each location in aspen-dominated mixedwood forests of north-central Alberta, 1992–93. Locations are distributed across four Natural Subregions (SR) and two Natural Regions (NR) A Carabidae (97.2% coverage) B Staphylinidae (97.6% coverage). Bars represent species richness (± 95% CI) and points represent the exponential of the Shannon diversity index (± 95% CI). For locations where there was significant interannual variation, the results for species richness are placed at the bottom of each bar, and results for diversity are placed in square brackets [] and are indicated above the point for that location.
The dominant Carabidae included: Pterostichus adstrictus Eschscholtz, 1823 (23.0% of overall catch), Pterostichus pensylvanicus LeConte, 1873 (20.8%), Calathus ingratus Dejean, 1828 (11.4%), and Platynus decentis (Say, 1823) (9.3%) comprising > 5.0% of total catch at every location. Agonum retractum LeConte, 1846 (13.9%) and Scaphinotus marginatus (Fischer von Waldheim, 1820) (8.2%) were abundant at four of six locations (Suppl. material
Many species were uncommonly collected over the course of the study; eleven species of Carabidae and 20 species of Staphylinidae were singletons or doubletons in our data set, and an additional eleven species of Carabidae and 23 of Staphylinidae were represented by < 15 specimens (Suppl. material
Natural Region significantly explained assemblage structure for carabids (F1,94 = 21.06, p = 0.006) and staphylinids (F1,94 = 29.42, p = 0.001), accounting for 17.4% and 23.0%, respectively, of variation in the data (Fig.
Principal components analysis of epigaeic beetle assemblages in aspen-dominated mixedwood forests in north-central Alberta, 1992–93 A Carabidae B Staphylinidae. Each symbol represents the catch from a single pitfall trap averaged over two years. Black symbols represent the Lower Foothills Natural Region; open symbols represent the Central Mixedwood Subregion, light grey symbols represent the Dry Mixedwood Subregion and the dark grey symbols represent the Lower Boreal Highlands Subregion of the Boreal Forest Natural Region. Ellipses represent the standard deviation around the centroid of points in each group and dashed polygons represent the final nodes of the multivariate regression tree (MRT) model. The inset shows the major MRT nodes grouping lines based on assemblage structure.
Multivariate regression tree analysis (MRT) of the carabid data by lines resulted in a tree with two significant splits (split 1: F1,82 = 71.2, p < 0.001; split 2: F1,82 = 25.7, p = 0.033) forming three clusters: group 1, which included the assemblages from the Hinton HIA and HIC lines and those from Slave Lake; group 2, which included the Hinton HIB line, and those from George Lake, Rose Creek, Lac la Biche Mature, and the O4 line at Lac la Biche; and group 3, which included the Peace River lines and the O2 line at Lac la Biche (Fig.
Eleven carabid and 17 staphylinid species were identified as significant indicator species (Table
Carabidae and Staphylinidae species identified as indicators of a particular Natural Region, Subregion, or combination.
Family | Carabid species | Natural Region (NR), Natural Subregion (SR), or combination | Indicator value* | p-value** |
---|---|---|---|---|
Carabidae | Calosoma frigidum Kirby, 1837 | Foothills NR | 0.826 | 0.007 |
Calathus advena (LeConte, 1846) | Foothills NR | 0.647 | 0.007 | |
Leistus ferruginosus Mannerheim, 1843 | Foothills NR | 0.638 | 0.007 | |
Pterostichus riparius (Dejean, 1828) | Foothills NR | 0.822 | 0.007 | |
Trechus chalybeus Dejean, 1831 | Foothills NR | 0.776 | 0.007 | |
Agonum retractum LeConte, 1846 | Boreal Forest NR | 0.975 | 0.007 | |
Agonum cupreum Dejean, 1831 | Dry Mixedwood SR | 0.490 | 0.043 | |
Harpalus fulvilabris Mannerheim, 1853 | Dry Mixedwood SR | 0.861 | 0.007 | |
Pterostichus melanarius (Illiger, 1798) | Dry Mixedwood SR | 0.813 | 0.007 | |
Synuchus impunctatus (Say, 1823) | Dry Mixedwood SR | 0.951 | 0.007 | |
Scaphinotus marginatus (F.v. Waldheim, 1820) | Central Mixedwood SR + Dry Mixedwood SR + Foothills NR | 0.919 | 0.007 | |
Staphylinidae | Micropeplus laticollis Mäklin, 1853 | Foothills NR | 0.527 | 0.034 |
Philonthus varians (Paykull, 1789) | Foothills NR | 0.615 | 0.024 | |
Quedius m. molochinoides Smetana, 1965 | Foothills NR | 0.737 | 0.014 | |
Tachinus frigidus Erichson, 1839 | Foothills NR | 0.948 | 0.014 | |
Tachinus quebecensis Robert, 1946 | Foothills NR | 0.770 | 0.014 | |
Tachyporus abdominalis (Fabricius, 1781) | Foothills NR | 0.577 | 0.034 | |
Lathrobium fauveli Duvivier, 1883 | Boreal Forest NR | 0.768 | 0.034 | |
Quedius l. labradorensis Smetana, 1965 | Boreal Forest NR | 0.930 | 0.014 | |
Tachinus fumipennis (Say, 1832) | Boreal Forest NR | 0.970 | 0.014 | |
Ischnosoma splendidum (Gravenhorst, 1806) | Lower Boreal Highlands SR | 0.893 | 0.014 | |
Mycetoporus americanus Erichson, 1839 | Lower Boreal Highlands SR | 0.819 | 0.014 | |
Philonthus flavibasis Casey, 1915 | Lower Boreal Highlands SR | 0.554 | 0.043 | |
Quedius brunnipennis Mannerheim, 1843 | Lower Boreal Highlands SR | 0.615 | 0.024 | |
Habrocerus schwarzi Horn, 1877 | Dry Mixedwood SR + Lower Boreal Highlands SR | 0.836 | 0.043 | |
Quedius fulvicollis (Stephens, 1833) | Foothills NR + Lower Boreal Highlands SR | 0.766 | 0.043 | |
Quedius velox Smetana, 1971 | Foothills NR + Lower Boreal Highlands SR | 0.860 | 0.014 | |
Tachinus elongatus Gyllenhal, 1810 | Foothills NR + Lower Boreal Highlands SR | 0.919 | 0.014 |
For carabids, the two climatic variables included in the environmental model, the number of frost-free days and the length of the normal dry period, together accounted for 27.7% of the total variation (Fig.
Redundancy analysis of epigaeic beetle assemblages in aspen-dominated mixedwood forests in Alberta, 1992–93 A Carabidae B Staphylinidae. Each symbol represents the catch from a single pitfall trap averaged over two years. Open symbols represent the Foothills Natural Region; the black symbols represent the Central Mixedwood Subregion, grey squares represent the Dry Mixedwood Subregion and the grey diamonds represent the Lower Boreal Highlands Subregion of the Boreal Forest Natural Region. Ellipses represent the standard deviation around the centroid of points in each group.
For staphylinids, the two climatic variables included in the environmental model were mean summer temperature and mean annual temperature, which together accounted for 32.0% of the total variance (Fig.
Overall β-diversity, including variation between locations, did not differ significantly between Natural Regions for carabids (F1,94 = 0.08, p = 0.893; Fig.
Boxplots of epigaeic beetle assemblage beta-diversity within regions, within locations and within lines for Boreal Forest (BF) and Foothills (FH) Natural Regions A Carabidae B Staphylinidae. Significant differences between Natural Regions are indicated by asterisks above the boxplots (** p < 0.01, *** p < 0.001).
Total variance was lowest between lines for both beetle groups (carabids: 20.1%; staphylinids: 9.9%; Fig.
Based on data from all 12 lines of traps in two age classes of stands at Lac la Biche, carabid and staphylinid assemblages differed significantly between mature and old stands (carabids: F1,70 = 19.47, p = 0.013; staphylinids: F1,70 = 8.51, p = 0.003; Fig.
For carabids, the final environmental model contained seven selected variables: graminoid cover, litter biomass, diversity of tall shrubs, southern aspect > 2°, slope, Salix spp. cover, and forb diversity. Collectively these accounted for 38.0% of the variance in carabid species composition at Lac la Biche (Fig.
Although there was a consistent pattern of higher overall β-diversity in old than mature stands at Lac la Biche, this pattern was not significant for carabids at any spatial scale (Locations: F1,70 = 0.33, p = 0.763; Stands: F1,70 = 0.93, p = 0.378; Lines: F1,70 = 2.11, p = 0.163; Fig.
Boxplots of beta-diversity of epigaeic beetle assemblages in aspen-dominated mixedwood forests at the Lac la Biche location, 1992–93 A Carabidae and B Staphylinidae. Beta diversity was calculated within locations, within stands, and within lines for mature and old stands. Significant differences between stand age classes are indicated by asterisks above the boxplots (* p < 0.05, ** p < 0.01, *** p < 0.001)
Overall, mean pitfall trap catches differed significantly between years for Carabidae (χ2 = 15.1, p < 0.001; 1992: 68.4 ± 2.9; 1993: 88.1 ± 4.7) but not for Staphylinidae (χ2 = 3.0, p = 0.08; 1992: 54.4 ± 3.0; 1993: 48.2 ± 2.7). Analysis of catches of both taxa showed significant interactions between years and locations (Carabidae: χ2 = 160.9, p < 0.001; Staphylinidae: χ2 = 78.0, p < 0.001). Carabid catches were significantly higher in 1992 than 1993 at Rose Creek but lower in 1992 than 1993 in both stand ages at Lac la Biche (Fig.
Epigaeic beetle composition differed significantly between years for many of the locations (Table
Interannual variation in Carabidae and Staphylinidae collected in 1992 and 1993. Comparison of faunal similarity between years is based on the standardized catch of beetles collected in 1992 and 1993 using the Bray-Curtis measure (= 1- dissimilarity value). Multivariate permutational ANOVA (999 permutations) used to test differences in similarity between trap catches across years partitioned by location. P-values = † (0.05–0.10), * (< 0.05), ** (< 0.01), *** (< 0.001).
Location | Carabidae | Staphylinidae | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Percent similarity-catch | Number of species collected only in 1992[1993] | Percent similarity-catch | Number of species collected only in 1992[1993] | |||||||||||
All species | Catch > 5 | Presence | All species | Catch > 5 | Presence | |||||||||
George Lake | 0.811 | 0.814 | 0.757 | * | 2[7] | 0.800 | † | 0.801 | † | 0.682 | ** | 6[8] | ||
Hinton | 0.763 | * | 0.764 | * | 0.875 | * | 3[1] | 0.594 | *** | 0.596 | *** | 0.762 | ** | 19[1] |
Lac la Biche-mature | 0.723 | *** | 0.723 | *** | 0.786 | 3[3] | 0.736 | *** | 0.739 | *** | 0.806 | *** | 7[6] | |
Lac la Biche-old | 0.557 | *** | 0.557 | *** | 0.857 | 2[3] | 0.682 | *** | 0.684 | *** | 0.845 | *** | 4[7] | |
Peace River | 0.857 | 0.870 | 0.690 | † | 8[1] | 0.507 | ** | 0.510 | ** | 0.818 | ** | 6[6] | ||
Rose Creek | 0.698 | ** | 0.699 | ** | 0.727 | 6[3] | 0.671 | *** | 0.674 | *** | 0.788 | 7[7] | ||
Slave Lake | 0.824 | 0.826 | 0.857 | 3[1] | 0.623 | *** | 0.625 | *** | 0.824 | * | 3[9] |
Application of PCA ordination to carabid and staphylinid data from each year separately (Fig.
Principal components analysis of epigaeic beetle assemblages in aspen-dominated mixedwood forests in north-central Alberta, 1992–93 A Carabidae, 1992 B Carabidae, 1993 C Staphylinidae, 1992 D Staphylinidae, 1993. Each point represents the catch from a single pitfall trap. Ellipses represent the standard deviation around the centroid of points in each group.
Although species richness of Staphylinidae was almost twice that of the Carabidae, both total catch and mean catch per trap was higher for the Carabidae. Mean catch, species richness, and species diversity differed among locations with no clear patterns for either taxon (Figs
In summary, carabid assemblages seem to respond to variation on larger spatial scales, while staphylinids respond to differences at finer spatial scales. A strong spatial trend was identified for Carabidae at Lac la Biche with distinct assemblages in eastern and western stands explaining much of the variance at the stand scale (Figs
Pitfall trap catch is strongly influenced by environmental variables such as temperature and precipitation, which affect activity (
It seems unlikely that this pattern of low catch and high diversity of carabids is characteristic of the Lower Boreal Highlands Subregion or northern forests. Extensive subsequent sampling of carabid assemblages in aspen forests at another locality (EMEND) in the same Subregion, ca. 55–60 km to the northwest of the stands sampled in the present study, did not show unusually low catch or high species richness/diversity (
Staphylinid diversity was highest in the three stands near Hinton in the Foothills Natural Region, likely reflecting a local mix of boreal, northern, and subalpine/montane elements (
The six carabid species caught most frequently in this study are generally dominant in forests across much of Canada, from Alberta eastward (
Differences between years, as documented above, can also affect interpretation of relationships among assemblages at different locations, the essence of understanding β-diversity. Of course, interannual variability, driven mainly by climactic variation, is one aspect of such differences; however, it is clear that even assemblages in more stable successional stages of forest, as we have studied here, can be expected to change progressively over time (
A significant question for conservation biologists interested in arthropod faunas is whether the structure of assemblages must be matched and maintained, or whether practices that maintain populations of all species in some combination are sufficient. Answering this question is complicated by the fact that arthropod species are differentially affected by climate change at the same time that other anthropogenic changes are also affecting natural habitats (
The most distinct and consistent spatial pattern revealed by our study is that assemblages of both carabids and staphylinids from the Lower Foothills Subregion (Hinton) differed greatly from most of those from Boreal Forest Subregions. As the single exception, carabid assemblages at Slave Lake were more similar to those at Hinton. The distinct nature of the Hinton assemblage is underscored by the high number (eleven) of indicator species for that location. There was also a major shift in dominance patterns among staphylinid species between Hinton and the boreal locations. Locations in the Boreal Forest Natural Region were dominated by T. fumipennis, whereas this species was largely replaced in dominance by T. elongatus, T. quebecensis Robert, 1946, T. vergatus Campbell, 1973, T. thruppi Hatch, 1957, and especially by T. frigidus, in the Lower Foothills (Suppl. material
Nonetheless, epigaeic assemblages of some stands near Hinton were more similar to those of boreal locations. The MRT analysis for staphylinids separated all three Hinton stands from those of the boreal locations at the first split in the MRT, indicating a high level of between-stand similarity in faunal structure. However, in contrast, the MRT for carabids resulted in assemblages from the HIA and HIC stands and Slave Lake stands separating together at the first split, while the HIB stand was grouped with some boreal locations in the second split. We note that the HIB stand was closer to a highly disturbed peri-urban area near the Athabasca River and at a much lower elevation (ca. 1000 m asl) than to the other two sites that had no urban or riverine influence (1160–1190 m asl). These differences are likely to be causally associated with presence of P. melanarius, a synanthropic introduced species (
The relationships of carabid and staphylinid faunas between Slave Lake and other locations differed as depicted in PCA ordinations. Because the two stands sampled near Slave Lake were in a transition zone between the Lower Foothills Subregion of the Foothills Natural Region and the Central Mixedwood Subregion of the Boreal Forest Natural Region (
The PCA also showed clearly that beetle assemblages varied between Natural Regions; however, Subregions within the Boreal Forest Natural Region did not consistently explain spatial differences among locations for either carabid or staphylinid assemblages. In addition, spatial patterns of association were substantially different for carabids and staphylinids. For carabids, assemblage structure at Peace River (Lower Boreal Highlands) was generally distinct from that at other locations, most of which grouped closely together irrespective of Subregion. Nonetheless, the MRT grouped assemblages from one old Lac la Biche stand (LBO2) with those from the Peace River stands. For staphylinids, however, MRT grouped assemblages at Peace River with most other boreal locations except the more southerly stands at Rose Creek and one stand at George Lake.
Separation in ordination space of the carabid assemblages from the Peace River stands from those of other boreal locations may in part be attributed to the absence of Synuchus impunctatus (Say, 1823) in trap catches from Peace River. This species was present, if uncommon (12–27 specimens), at the other boreal locations. Furthermore, the capture of three specimens of Pterostichus brevicornis (Kirby, 1837), a species with more northerly distribution, at Peace River contrasts with its absence in catches at other boreal locations. However, because of the very low overall carabid catch at Peace River compared to other boreal locations, small changes in relative catch of different species could result in relatively large changes in assemblage structure compared to locations with much larger catches. Thus, some of the unique structure of Peace River carabid assemblages could be simply a statistical artifact of low catch. Nonetheless, absence of S. impunctatus in samples from stands near Peace River seems characteristic for this Subregion as extensive sampling of mature pyrogenic aspen stands elsewhere in this Subregion between 1998 and 2013 yielded very few individuals, and frequently none, in most sample years (
In the RDAs, climate variables explained >50% of the constrained variation for both carabids and staphylinids across regions, more than that explained by vegetation and other site characteristics. Separation of assemblages from the two Natural Regions appears to reflect the importance of climate as foothills environments are generally cooler and wetter than boreal locations. Both soil moisture content and precipitation are correlated with temperature and have been reported as a prime driver of epigaeic beetle diversity in forests by many authors (e.g.,
Correlation between coverage by ferns and tall shrub species such as Salix, Alnus and Prunus with epigaeic beetle assemblage structure suggests that both groups respond to similar environmental parameters. Several studies have shown that structure of both epigaeic beetle assemblages and plant communities (trees and understory plants) are influenced by environmental factors such as soil moisture and nutrients (
Variance partitioning showed that environmental variables explained the largest fraction of the total variance in both carabid and staphylinid assemblage structure at the largest spatial scale (i.e., between locations). Furthermore, explanatory power of these variables decreased as spatial scale decreased, irrespective of total variance at each spatial scale. Of course, this is expected because at smaller spatial scales environments become more similar; however, it may also underscore the importance of unmeasured environmental variables that are generally not included in broad forest classification schemes but may be increasingly influential in explaining variance in beetle assemblages as spatial scale decreases. Microhabitats such as amount and quality of coarse woody debris, mammal scat, mushrooms, conifer needle beds, squirrel middens, persistent pools of standing water, mineral soil exposure, etc. can influence species and catch of epigaeic beetles (
The fact that β-diversity was higher for staphylinids than for carabids at every spatial scale reflects the fact that α-diversity was about twice as high for staphylinids at all spatial scales. The two exceptions to this general pattern are the within-region and within-locality β-diversity for the Foothills Natural Region, which were about the same for both beetle families (note: within-region and within-locality β-diversity is identical because there was only one location sampled (Hinton) in this Region). As our sampling covered a much larger geographic area and diversity of stand types (i.e., three Subregions) in the Boreal Forest than in the Lower Foothills, we expected that β-diversity would be higher in the Boreal Forest. This was the case for staphylinids but not carabids. We interpret this to reflect the fact that the carabid assemblage in the HIB stand at Hinton differed significantly from the other two stands, as discussed above, and this is the principal explanation for the elevated β-diversity that we observed. This also explains why within-location β-diversity for carabids was significantly higher at Hinton than at other locations that covered similar areas, while within-location β-diversity for staphylinids did not differ between regions. The similarity of within-line β-diversity among regions suggests that microhabitat variability at this scale was about the same in each region.
The expanded sampling effort in Lac la Biche stands (2–4 trap lines per stand) in comparison to other locations (1 line per stand) supported investigation of whether never-cut primeval forests (> 120 years old) harboured unique biodiversity in comparison to anticipated rotation-age forests (mature stands ca. 40–60 years;
Although sampling location and stand age were identified as primary drivers of epigaeic beetle assemblage structure at Lac la Biche, these factors affected the two beetle families differently. Carabid assemblages clearly separated into western and eastern groups, and within the western group there was a more distinct influence of stand age than in the eastern group. In contrast, staphylinid assemblage structure was most clearly influenced by stand age but apparently unaffected by stand location. While the influence of stand age has been previously discussed (
The RDAs indicated that differences in microclimate associated with slope, aspect and vegetation were correlated with structure of epigaeic beetle assemblages at the local and regional levels, and these factors have been reported as important for structuring epigaeic beetle communities in other studies from Alberta (
The apparently contradictory association of graminoids with old stands likely reflects operation of gap dynamic processes that introduce patchy openings in late successional stands (
Nonetheless, variance partitioning in epigaeic beetle assemblages at Lac la Biche revealed patterns similar to those evident at the regional level, suggesting that variance patterns can be generalized across spatial scales. Clearly environmental variables such as climate, plant community parameters, aspect and slope that act primarily at medium to large scales, where they are most easily measured, and they explain the majority of variance at the largest spatial scales for both families. However, variables that impact beetle populations at smaller (microhabitat) scales and are more difficult to measure have the greater influence on beetle catches at the smallest scales. This is especially true for staphylinids that seem to be more attuned to microsite differences than carabids, which are usually wide-ranging, generalist predators. Thus, both our regional and local-level data sets suggest that composition of carabid beetle assemblages is governed more by ‘coarse’ scale influences that can be considered in forest management. In contrast, staphylinid assemblages are governed more by ‘fine’ scale influences that will be perhaps more challenging to manage.
The two patterns of decreasing β-diversity with decreasing spatial scale and consistently higher β-diversity for staphylinids than carabids at each spatial scale was consistent across both the regional and local datasets. However, the β-diversity pattern with respect to stand age class was less consistent, as age class clearly affected β-diversity of staphylinids but was not apparent in our carabid data. The higher β-diversity observed in old stands compared to mature stands at all three spatial scales suggests that old stands have a higher level of microhabitat diversity that is important to many staphylinid species (
Although aspen-mixedwood forests are similar with respect to dominant vegetation structure across Natural Subregions in Alberta, our study has clearly shown that epigaeic beetle assemblages differ across this extensive landscape. It is also clear that these two families of beetles respond to various environmental influences at different spatial scales. This is significant for development of forest management strategies aimed to conserve biodiversity. For example, a similar study in central Europe comparing effects of spatial scale on biodiversity in managed forests concluded that results from a single region should be generalized cautiously because variation in site history and composition of epigaeic beetle communities may result in different responses to similar forest management strategies across a broad landscape (
β-diversity is scale dependent and can be influenced by the number and distribution of sampling sites. We have shown that considerable variation exists, even at local scales (e.g., Carabidae at Hinton), and that effective study of these patterns likely requires sampling that considers resources specific to the taxon of interest (e.g., mushrooms or dead wood for Staphylinidae). Understanding of small-scale variability is elusive, and depends to some extent on stand age and successional trajectory (
Our study also suggests that subtle climatic variation plays a significant role in the distribution of epigaeic beetles across large mixedwood landscapes in Alberta. Therefore, systems such as the Natural Regions classification are appropriate to guide development of conservation plans for these groups. Forest classification schemes at even smaller spatial scales, such as subregions, may be helpful for carabids, but will not likely work as well for staphylinids.
Further study increasing the number of replicate stands across regions and taking account of other landscape characteristics such as aspect and slope as well as land use patterns and history may provide additional information useful for faunal conservation. However, with climate change a certainty and multiple land uses occurring across these forested landscapes, we can expect redistribution of species across forested sites and resulting changes in patterns of assemblage structure. Thus, an ongoing challenge will be to differentiate effects of disturbance and land use activities that can be managed from the effects of global change. This will require study and understanding of species in relation to baseline data such as that provided here.
Through his important work and discussion, Terry Erwin expanded our attention from describing assemblages to considering how they change in space as an important element of understanding biodiversity.
We thank Hector Cárcamo, Tim Spanton, Jari Niemelä, Karen Cryer, Nora Berg, Kevin Sytsma, Tom Clarke, Peter Shipley from the CFS and the University of Alberta for field assistance, specimen sorting and insect identification; Florence Niemi and Bob Wynes from Daishawa-Marubeni International for help with site selection near Peace River; Norbert Raphael from Alberta Parks for issuing research permits for Lakeland Provincial Park; and Brad Stelfox, Philip Lee and Susan Crites from the Alberta Environmental Centre for assistance with site selection near Lac la Biche. We have tried to work at the species level as Erwin advocated in ecological studies. Even in north temperate areas this is impossible without help of taxonomic specialists. Thus, we thank Danny Shpeley and George Ball (Carabidae) and Margaret Thayer (Staphylinidae) for assistance with identification of difficult specimens. Funding for this study was graciously provided by Natural Resources Canada – Canadian Forest Service and a Natural Sciences and Engineering Research Discovery Grant to JRS.
Supplementary tables
Data type: species data
Explanation note: Table S1. Raw catch of Carabidae and Staphylinidae species collected from boreal aspen-dominated mixedwood forests in north-central Alberta, 1992–1993, and number of traps days per line. Species catch are summed across years of collection. Nomenclature is based on