Northeastern Naturalist B135 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 Winter Movements of Sitta canadensis L. (Red-breasted Nuthatch) in New England and Beyond: A Multiple-scale Analysis W. Herbert Wilson Jr.1,* and Bets Brown1 Abstract - We analyzed 55 years of abundance data (1960–2014) for Sitta canadensis (Redbreasted Nuthatch) to seek patterns of winter irruptions on temporal and spatial scales. This species shows an erratic pattern of irruption into southerly areas from its northern breeding areas. Irruptions show a broad geographic synchrony. At the narrower level of the state or province and even more so at the level of individual counts, correlations of abundance in adjacent areas become weaker. The abundance of irruptive birds is best considered a mosaic. At the regional scale, correlations of Red-breasted Nuthatch abundance with irruptive northern finches that also depend on conifer seeds are weak to absent. The data suggest that birds irrupt because of failure of conifer seed production on the breeding grounds, not because the birds are seeking masting conifer stands to the south. Introduction Seasonal migrations between breeding and wintering grounds characterize a large number of bird species in the Northern Hemisphere. A smaller group of species migrates south in movements called irruptions in some but not all years. In this paper, we define an irruption as a significant fall to early winter invasion of birds with distributions centered in the high-latitudes to more southerly areas where the species do not regularly winter (Bock and Lepthien 1972, 1976; Lack 1954). The best known of these irruptive species in North America are a group of fringillids called the northern finches, including Acanthis flammea (L.) (Common Redpoll), A. hornemanni (Holbøll) (Hoary Redpoll), Spinus spinus (L.) (Pine Siskin), Haemorhous purpureus (Gmelin) (Purple Finch), Pinicola enucleator (L.) (Pine Grosbeak), Coccothraustes vespertinus (W. Cooper) (Evening Grosbeak), Loxia leucoptera Gmelin (White-winged Crossbill), and L. curvirostra L. (Red Crossbill) (Bock and Lepthien 1976, Erickson 1970, Evans 1966, Reinikainen 1937). All of these species rely on conifer seeds for much of their food (Bock and Lepthien 1976) and their irruptions are often synchronous (Bock and Lepthien 1972, 1976). Sitta canadensis L. (Red-breasted Nuthatch) is also an irruptive species. Redbreasted Nuthatches frequent coniferous forests (Adams and Morrison 1993, Ball 1947, Wydemerer 1933) and breed across most of southern Canada and the northern tier of the US, with populations extending southward along the Appalachians to North Carolina and the Rocky Mountains to Colorado. The diet of Red-breasted 1Department of Biology, Colby College, Waterville, ME 04901. *Corresponding author - whwilson@colby.edu. Manuscript Editor: Heather York Winter Ecology: Insights from Biology and History 2017 Northeastern Naturalist 24(Special Issue 7):B135–B146 Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B136 Vol. 24, Special Issue 7 Nuthatches consists mostly of arthropods during the breeding season and conifer seeds during the winter (Ghambalor and Martin 1999). Red-breasted Nuthatches cache seeds for later use during the winter. The conventional explanation for irruptions is variability in seed production by conifers or other trees whose seeds provide a major food source for the irruptive birds (Formozov 1960, Lack 1954, Newton 1970, Smith 1986). During mast years, birds are predicted to remain on their breeding grounds through the winter, but are forced southward when seed production is low. We currently lack continent-wide data on conifer cone production to test this hypothesis directly. However, we can employ an indirect test by looking for synchrony between Red-breasted Nuthatches and northern finches. Though not closely related, Red-breasted Nuthatches and finches depend on conifer seeds for winter sustenance. If levels of conifer seed production drive irruptive movements, then the finches and the nuthatch should show similar patterns. We also consider 2 alternative hypotheses: (1) birds periodically move south in search of masting conifers, irrespective of local cone abundance, and (2) birds are competitively forced south by population increases following particularly successful nesting seasons. In this paper, we use data from the National Audubon Society Christmas Bird Counts (CBCs) to determine patterns of irruption in Red-breasted Nuthatches. We examine variability broadly from a continental perspective as well as smaller geographic units (state- and province-level down to the level of individual count-circles). Finally, we examine the strength of correlations of abundance of Red-breasted Nuthatches with several species of irruptive finche s. Materials and Methods The primary source for the data used in this paper was the National Audubon Christmas Bird database. A CBC is conducted annually on a single day sometime between mid-December and early January by thousands of observation teams who each count birds within a fixed 460-km2 (15 miles in diameter) circle. We downloaded the state and province data from the National Audubon Society website (https://www.audubon.org/conservation/science/christmas-bird-count). These data included the mean abundance (measured as number of birds per party-hour) for each state or province of all the CBCs conducted in a particular year. We requested and received the count-level data for all counts on which Red-breasted Nuthatches appeared since the beginning of the program in 1900. Analysis of the number of CBC count circles included each year indicated a dramatic rise beginning around 1960 and continuing to the present. We therefore excluded data from earlier than 1960 because of limited sample size for some states and provinces. CBCs have the advantage of sampling a standardized area. However, the number of observers varies greatly among counts. Therefore, we adopted the metric used by National Audubon Society biologists—number per party-hour. This method standardizes the abundance to account for differences in effort. CBC data do not take into account detectability of different species, but this shortcoming is obviated Northeastern Naturalist B137 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 by our consideration of a single species. The early-winter timing of CBCs may not capture the full irruption in a given year. To examine the claim that Red-breasted Nuthatches show a biennial pattern of irruption, we graphed numbers of birds per party-hour against year. For our analyses, we included 3 states within the breeding range and 3 states in which Redbreasted Nuthatches only occur as winter visitors. We employed extensive exploratory graphical analysis in this research. We used the choroplethr package in R (R Core Team 2016) to make choropleths (maps where areas are shaded in proportion of a measured variable) of various data for 59 administrative regions (the lower 48 states, the District of Columbia, the 10 provinces along the southern tier of Canada). We use the term “regions” in this paper to denote US states and Canadian provinces. We created choropleth maps to show the region-level mean abundance from the period 1960 until 2014. The coefficient of variation (CV) for each region was also mapped with a choropleth. To examine the effect of distance at the regional level, we performed correlation analyses on all possible combinations of the 59 regions. We then chose 8 regions for graphical analysis. Using a choropleth, we mapped the correlation coefficient (Pearson’s r) for a focal region with all other regions. At a finer scale, we tested for similarities of abundance at the count-circle level. We chose 8 count circles in 8 different regions as focal areas. We chose these circles a priori based on the length of the count history and the number of other counts within a 250-km radius. We employed a geosphere in R (R Core Team 2016) to determine the distance between the focal circle and all other count circles. We filtered the data to only include those circles within 300 km of the focal circle. For those circles, we used the rcorr package in R (R Core Team 2016) to determine the Pearson-r and the P-values for each correlation between the focal circle and the other circles in the 300- km radius. We prepared histograms of distance and indicated significance (P < 0.05) or lack of significance by color for each count-circle considered. To test for synchrony of irruptions at the regional level, we performed correlation analyses using the rcorr package in R (R Core Team 2016) for Red-breasted Nuthatches with 3 fringillids: Common Redpoll, Pine Siskin, and Purple Finch. We report data for the 10 northern regions where Red-breasted Nuthatches may be found year-round. Results Figure 1 presents annual mean abundance (number per party-hour) for 6 regions. The graphs do not show a consistent saw-tooth pattern one would expect for a true biennial pattern. The expected consecutive abundance alternating above and below the grand mean does not consistently appear. Indiana and Ohio, areas south of the breeding range, show a bit more evidence of biennial alternation of highs and lows. However, some peaks occur in odd-numbered years and others in even-numbered years, forcing us to reject the hypothesis of biennial irruptions. The highest number per party-hour values occur in the western provinces, Washington, Montana, and the 3 three northern New England states (Fig. 2A). As Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B138 Vol. 24, Special Issue 7 expected, the lowest abundances occur in the areas most distant from breeding areas along the southern tier of states. Figure 1. Yearly abundance of Red-breasted Nuthatches, measured as number per partyhour for 6 states. The mean number per party-hour for each state is represented by a horizontal red line. Montana (MT), Minnesota (MN), Maine (ME), and Colorado (CO) host breeding populations of Red-breasted Nuthatches. Indiana (IN) and Ohio (OH) are south of the breeding range. Northeastern Naturalist B139 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 As a measure of variation within a region among years, we present the CV for each of the 59 regions (Fig. 2B). We expected the relatively high CV values in the southern states where irruptions are rare. More surprising is the high CV for Newfoundland, which has abundant coniferous forests. Most of the regions with the highest densities showed relatively low CVs (Montana, British Columbia, Maine, New Hampshire, Vermont). When we treated each region as a separate focal area, correlations of abundance with all other areas generally indicated a strong regional pattern. A given region was more likely to be correlated with adjacent regions compared to more distant regions. We present choropleths for 8 states to document this pattern. Maine shows a strong Figure 2. Choropleth maps showing (A) the abundance of Red-breasted Nuthatches measured as number per party-hour in the lower 48 states and the 9 southern Canadian provinces, and (B) the variability in abundance (measured as the coefficient of variation) in the same regions. Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B140 Vol. 24, Special Issue 7 correlation of yearly density with New Hampshire and Nova Scotia, but not Vermont or Quebec (Fig. 3A). Quebec shows a more consistent pattern with high correlations with New Brunswick, Nova Scotia, and Ontario (Fig 3B). Tennessee, well south of the Red-breasted Nuthatch’s breeding range, shows a regional signature with high correlations in adjoining Alabama, North Carolina, Virginia, and nearby West Virginia (Fig. 3C). Similarly, Minnesota shows a regional signature with abundances strongly correlated with Wisconsin, Ontario, and Michigan (Fig. 3D). Four western regions show the same evidence of regional variation (Fig. 4). Alberta shows strong correlations with the 3 other western provinces of British Columbia, Saskatchewan, and Manitoba (Fig. 4A). North Dakota and Montana show strong positive correlations with the Alberta abundances. The annual abundances in Montana are strongly correlated along the northern tier of regions, surprisingly strongly with Wisconsin, Michigan, Ontario, Quebec, and New Brunswick (Fig. 4B). A strong northwestern pattern of similar abundances is manifested in the Oregon- centered choropleth (Fig. 4C) where Washington, British Columbia, and Idaho are strongly correlated, and Alberta, Saskatchewan, and Montana are only slightly less strongly correlated. Colorado shows a more restrictive pattern, with a strong correlation with Utah and a fairly strong correlation with New Mexico (Fig. 4D). Other adjoining states show weaker correlations. Examination of correlations of number per party-hour among individual CBC circles allows for a more granular comparison of variability (Fig. 5). For distances among counts separated by no more than 300 km, we found surprisingly poor significant correlations with a central focal circle. The number of significant correlations did not fall off with increasing distance from the focal circle. For each of the 4 focal circles in Saskatchewan (Fig. 5A), Maine (Fig. 5B), Oregon (Fig. 5C), and Montana (Fig. 5D), most of the correlations were not signifi cant. Table 1 presents correlation analyses testing for synchrony in the abundance of Red-breasted Nuthatches with 3 species of irruptive finches that also rely on small cones from conifer trees. Common Redpoll abundance was significantly correlated with Red-breasted Nuthatch abundance in Washington and in New York. None of the other regions showed a significant correlation between these 2 species. We Table 1. Regional-level correlation analysis of Red-breasted Nuthatch abundance (number per partyhour) with the abundance of Common Redpoll, Pine Siskin, and Purple Finch from the period 1960 to 2014. * denotes r-values that are significant at the P < 0.01 level. Region Common Redpoll Pine Siskin Purple Finch ME 0.031 0.154 -0.086 NH 0.151 0.001 0.052 VT 0.081 0.249 0.619* NY 0.542* 0.134 0.060 MI 0.023 0.275 -0.322 MN -0.157 0.200 0.039 MT 0.023 0.275 -0.323 WA 0.490* 0.124 0.010 OR 0.177 0.120 -0.078 Northeastern Naturalist B141 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 Figure 3. Choropleth maps showing correlations in abundance with 4 representative focal centers from eastern and central North America: (A) Maine, (B) Quebec, (C) Tennessee, and (D) Minnesota. Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B142 Vol. 24, Special Issue 7 Figure 4. Choropleth maps showing correlations in abundance with 4 representative focal centers from western North America: (A) Alberta, (B) Montana, (C) Oregon, and (D) Colorado. Northeastern Naturalist B143 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 found no significant correlations for nuthatch abundance with that of Pine Siskin, and only Vermont showed a significant positive correlation for nuthatch and Purple Finch abundance. Discussion Irruptive migrations are common, though unpredictable, behaviors of a number of northerly species in North America and Eurasia (Ball 1947, Lack 1954). Despite extensive documentation of these movement patterns, our understanding of the drivers of the phenomenon is limited. Attempts to understand a phenomenon that occurs on the level of continents or hemispheres is a daunting task. Nevertheless, 3 hypotheses have emerged to explain irruptive migrations. First, failure of food supplies on the northern breeding grounds has been claimed to provide the impetus for irruptions from the breeding grounds into more southerly areas (Bock and Lepthien 1972, 1976; Lack 1954). For species like finches and the Figure 5. Correlations of Red-breasted Nuthatch abundance among proximate CBC circles (within 300 km of the focal circle) using a focal circle in (A) Saskatchewan, (B) Maine, (C) Oregon, and (D) Montana. For each CBC circle, the teal color denotes a significant positive correlation with the focal circle and the orange color denotes a non-significant correlation with the focal circle. The distance from the focal circle is represented on the x-axis for each circle in the analysis. Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B144 Vol. 24, Special Issue 7 Red-breasted Nuthatch, masting in conifers on the breeding grounds could certainly impel these birds to forgo a southerly migration during a winter of relative plenty. Mast years are typically followed by 1 or more years of lowered cone-production, which could force cone-dependent birds to migrate. Although this explanation seems reasonable, we lack information on cone production in the northern forests. Furthermore, we do not know the geographic scale at which masting occurs for most northerly areas. Bock and Lepthien (1976) provided gross measurements of cone crops, showing that irruptions were more prevalent when cone crops in the north were poor. The best data come from Widrlechner and Dragula (1984), who used foresters’ measurements of commercially important conifers in California, Oregon, and Washington to show that Red-breasted Nuthatches appeared less frequently in California when cone crops were good in Oregon and Washington. Second, some researchers believe that the promise of masting conifers south of the normal range can induce birds to move (Lack 1954). We argue that this hypothesis is weak because it presumes omniscience on the part of the birds. To be sure, irruptive birds may take advantage of mast during an irruption. However, we believe the birds opportunistically take advantage of those food supplies but are likely driven south to wander until sufficient food can be found. Bagg (1969) describes a correlation between mast years in Ontario and irruptions of Poecile atricapillus (L.) (Black-capped Chickadee) into central Maine. The link between these 2 events requires that the incidence of masting be synchronous from Maine to Ontario. Similarly, James (1967) noted few irruptive birds in Arkansas during a mast year in Newfoundland. The final hypothesis is that birds irrupt from their breeding grounds after particularly successful reproductive seasons. We simply lack the data on reproductive success at higher latitudes to test this hypothesis now. Examination of the scale of irruptions provides some insight into the driving forces of such movement. For the Red-breasted Nuthatch, Bombycilla garrulus (L.) (Bohemian Waxwing), and a suite of northern finches, Bock and Lepthien (1976) demonstrated broad-scale geographic patterns of irruption followed by years where few irruptives appeared. These irruptions are less predictable at smaller scales (Figs. 3, 4). Using 55 years of CBC data, we showed that Red-breasted Nuthatch abundance is usually strongly correlated in adjacent states or provinces but is often quite different from regions that are farther away. For some comparisons, even adjacent regions are poorly correlated (e.g., Maine and Quebec, Colorado and Wyoming). Red-breasted Nuthatches readily accept supplemental food at bird feeders (Wilson 1999, 2002; Yunick 1982); thus, differences in bird feeding in proximate count circles may contribute to the poor correlations observed. Correlations are even weaker at even more-localized scales (Fig. 5). This analysis implies that irruptive species are patchy at small scales; correlations between small areas, such as the 460-km2 of a CBC circle, are weak and insignificant for most comparison s. We regard the distribution of Red-breasted Nuthatches as a mosaic of areas of high and low abundance when viewed at a fine scale, but as a more even when observed at larger scales. The count-level CBC data strongly indicate that Northeastern Naturalist B145 W.H. Wilson Jr. and B. Brown 2017 Vol. 24, Special Issue 7 Red-breasted Nuthatch abundance within focal regions is not uniform in any given year (Fig. 4). We believe that the best explanation for irruptive migration is food limitation on the breeding ground. Depending on the amount of food available, smaller or larger fractions of a population may be forced to irrupt southward. Those birds continue south until they find sufficient food, which is often associated with local masting. We do not believe the birds are attracted to a particular area because of likely masting; rather, the birds wander until they find food. Masting forests south of the breeding range may therefore never be used by irrupting birds. Furthermore, heterospecific birds may wander in different directions, explaining the lack of strong correlations among different species at the regional level (Table 1). Although we provide a hypothesis to explain the irruptive behavior of Redbreasted Nuthatches, more data and efforts are needed to fully understand the direction, scale, and impetus for these irruptions. Tracking conifer cone production across the continent and geo-tagging or satellite-tagging birds will be necessary to offer a robust test of our hypothesis. Acknowledgments We thank Kathy Dale, Director of Citizen Science at the National Audubon Society. Liam O’Brien and Manuel Gimond provided statistical advice. Thanks also to the thousands of participants in the Christmas Bird Count program, whose collective efforts have made the CBC database such a powerful source of information on changing bird populations. Literature Cited Adams, E.M., and M.L. Morrison. 1993. Effects of forest-stand structure and composition on Red-breasted Nuthatches and Brown Creepers. Journal of Wildlife Management 57:616–629. Bagg, A.M. 1969. A summary of the fall migration season, 1968, with special attention to the movements of Black-capped Chickadees. Audubon Field Notes 23:4–12. Ball, S.C. 1947. Migration of Red-breasted Nuthatches in Gaspé. Ecological Monographs 17:501–533. Bock, C.E., and L.W. Lepthien. 1972. Winter eruptions of Red-breasted Nuthatches in North America: 1950–1970. American Birds 26:558–561 Bock, C.E., and L.W. Lepthien. 1976. Synchronous eruptions of boreal seed-eating birds. American Naturalist 110:559–571. Erickson, K. 1970. Ecology of the eruption and wintering of Fennoscandian Redpolls (Carduelis flammea coll.). Annales Zoologica Fennici 7:273–282. Evans, P.R. 1966. Autumn movements, moult, and measurement of the Lesser Redpoll, Carduelis flammea. Ibis 108:183–216. Formozov, A.N. 1960. La production de grains dans les forêts de conifers de la taiga de l’U.S.S.R. et l’envahissement de l’Europe occidentale par certaines espèces d’oiseaux. Proceedings of the XII International Ornithological Congress. 1960:216–229. Ghambalor, C.K., and T.E. Martin. 1999. Red-breasted Nuthatch (Sitta canadensis). No. 459, In A. Poole and F. Gill (Eds.). The Birds of North America. The Birds of North America, Inc., Philadelphia, PA. James, F.C. 1967. Central-southern regional report. Audubon Field Notes 21:426–430. Northeastern Naturalist W.H. Wilson Jr. and B. Brown 2017 B146 Vol. 24, Special Issue 7 Lack, D. 1954. The Natural Regulation of Animal Numbers. Clarendon Press, Oxford, UK. 343 pp. Newton, I. 1970. Irruptions of crossbills in Europe. Pp. 337–357, In A. Watson (Ed.). Animal Populations in Relation to their Food Resources. Blackwell Press, Oxford, UK. 474 pp. R Core Team. 2016. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at https://www.Rproject. org/. Reinikainen, A. 1937. The irregular migrations of the Crossbill, Loxia c. curvirostra, and their relation to the cone crop of the conifers. Ornis Fennica 14:55–64. Smith, K.G. 1986. Winter population dynamics of three species of mast-eating birds in the eastern United States. Wilson Bulletin 98:407–418. Widrlechner, M.P., and S.K. Dragula. 1984. Relation of cone-crop size to irruptions of four seed-eating birds in California. American Birds 38:840–846. Wilson, W.H., Jr. 1999. Bird feeding and irruptions of northern finches: Are migrations short-stopped? North American Bird Bander 24:113–121. Wilson, W.H., Jr. 2002. Population and individual responses of Red-breasted Nuthatches (Sitta canadensis) to supplemental food in central Maine. North American Bird Bander 27:59–54. Wydemeyer, W. 1933. Zonal range of the Red-breasted Nuthatch in northwestern Montana. Condor 35:32–33. Yunick, R.P. 1982. Some factors influencing feeder selection by Black-capped Chickadees and Red-breasted Nuthatches. North American Bird Bander 7:20–23.