(LONICERA MAACKII) COMPARED to OTHER WOODY SPECIES By

Home , Acer rubrum, Liquidambar, North American beaver, Nymphaea, Quercus rubra

ABSTRACT

BEAVER (CASTOR CANADENSIS) ELECTIVITY FOR AMUR HONEYSUCKLE (LONICERA MAACKII) COMPARED TO OTHER WOODY SPECIES

by Janet L. Deardorff

The North American beaver is a keystone riparian obligate which creates and maintains riparian areas by building dams. In much of the eastern U.S., invasive shrubs are common in riparian zones, but we do not know if beavers promote or inhibit these invasions. I investigated whether beavers use the invasive shrub, Amur honeysuckle (Lonicera maackii), preferentially compared to other woody species and the causes of differences in L. maackii electivity among sites. At eight sites, I identified woody stems on transects, recording stem diameter, distance to the water’s edge, and whether the stem was cut by beaver. To determine predictors of cutting by beaver, I conducted binomial generalized regressions, using distance from the water’s edge, diameter, and plant genus as fixed factors and site as a random factor. To quantify beaver preference, I calculated an electivity index (Ei) for each genus at each site. Lonicera maackii was only preferred at two of the eight sites though it comprised 41% of the total cut stems. Stems that were closer to the water’s edge and with a smaller diameter had a higher probability of being cut. Among sites, L. maackii electivity was negatively associated with the density of stems of preferred genera.

BEAVER (CASTOR CANADENSIS) ELECTIVITY FOR AMUR HONEYSUCKLE (LONICERA MAACKII) COMPARED TO OTHER WOODY SPECIES

A Thesis

Submitted to the

Faculty of Miami University

in partial fulfillment of

the requirements for the degree of

Master of Science

Janet L. Deardorff

Miami University

Oxford, Ohio

2019

Advisor: Dr. David Gorchov

Reader: Dr. Susan Hoffman

Reader: Dr. Bartosz Grudzinski

This thesis titled

BEAVER (CASTOR CANADENSIS) ELECTIVITY FOR AMUR HONEYSUCKLE (LONICERA MAACKII) COMPARED TO OTHER WOODY SPECIES

Janet L. Deardorff

has been approved for publication by

The College of Arts and Sciences

and

Department of Biology

______Dr. David Gorchov

______Dr. Susan Hoffman

______Dr. Bartosz Grudzinski

Table of Contents LIST OF TABLES……………………………………………………………………………....IV LIST OF FIGURES……………………………………………………………………………....V LIST OF APPENDEXES………………………………………………………………………..VI ACKNOWLEDGEMENTS…………………………………………………………………….VII INTRODUCTION………………………………………………………………………………...1 Beaver Foraging Behavior………………………………………………………………...1 Herbivore/Plant Interactions………………………………………………………………2 Amur honeysuckle, Lonicera maackii…………………………………………………….2 Beaver Abundance in Ohio……………………………………………………………...... 3 Potential Importance of Beaver-Lonicera Interactions in Riparian Zones……………...... 3 Study Questions…………………………………………………………………………...3 METHODS……………………………………………………………………………………...... 4 Transect Configuration……………………………………………………………………4 Woody Plant Sampling……………………………………………………………………5 Canopy Openness………………………………………………………………………….5 Beaver Residency Time…………………………………………………………………...5 Density of Preferred Stems……………………………………………………………...... 5 Photos of Beavers Cutting………………………………………………………………...6 STATISTICAL ANALYSIS……………………………………………………………………...6 Probability of a Stem Being Cut………………………..…………………………………6 Electivity…………………………………………………………………………………..6 Linear Regressions……………………………………………………………………...... 7 RESULTS…………………………………………………………………………………………7 Factors Influencing Stem Cutting…………………………………………………………7 Electivities…………………………………………………………………………………8 Differences in L. maackii Electivity Among Sites………………………………………..8 DISCUSSION……………………………………………………………………………………..8 Difference Among Sites………………………………………………………………….10 Few Cut Stems in Spring Census………………………………………………………...10 Future Research………………………………………………………………………….11 CONCLUSION…………………………………………………………………………………..11 REFERENCES……...…………………………………………………………………………...13 TABLES…………………………………………………………………………………………17 FIGURES………………………………………………………………………………………...23 APPENDIX 1 ……………………………………………………………………………………37 APPENDIX 2…………………………………………………………………………………….41 APPENDIX 3…………………………………………………………………………………….48

iii

List of Tables

Table 1. The location, city, beaver residency time, management entity, aquatic system, and transect type, and field notes from spring/summer 2018 at each site. ….……………………….17

Table 2. Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut, using stems from all genera…………………...18 Table 3. Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut using stems from genera found in five or more sites……………………………………...……………………………………………………….19 Table 4. Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut using stems from L. maackii…………………...20 Table 5. The number of browsed stems for each genus and each diameter class at each site during the fall census……...……………………………………………………………………………..21

Table 6. The number of cut and browsed stems during the spring re-census in each genus and each diameter class.………………..………………………………………………………….....22

List of Figures

Figure 1. First survey of beaver in Ohio 1949 after extermination in 1830. The author inferred beaver colonies in the northwest came from Michigan and the eastern colonies came from Pennsylvania. Figure from Chapman (1947)……………………………..……………………...23

Figure 2. Map describing the distribution and abundance of beavers in Ohio in 2012. Image taken from http://wildlife.ohiodnr.gov/species-and-habitats/species-guide-index/mammals/beaver ………………………………………………………………………………………...………….24

Figure 3. Beaver population estimates in Ohio from 1980-2011. Image taken from http://wildlife. ohiodnr.gov/species-and-habitats/species-guide-index/mammals/beaver ..………………...…...25

Figure 4. Locations of study sites in southwest Ohio…………………………………………....26

Figure 5. Schematic of transect designs………………………………………………………………….27

Figure 6. Photos taken during summer 2018 from Browning (Model BTC-5HDE) trail cameras placed at Bachelor Pond (Oxford, Oh)…………………………………………………………..28

Figure 7. Barplot of proportion of stems cut out of all stems within each distance class………..29

Figure 8. Barplot of proportion of stems cut out of all stems within each diameter class………30

Figure 9. Barplot of proportion of L. maackii stems cut out of all L. maackii stems within each distance class……………………………………………………………………………………..31

Figure 10. Barplot of proportion of L. maackii stems cut out of all L. maackii stems within each diameter class…………………………………………………………………………………….32

Figure 11. Barplot of proportion of cut stems belonging to each genus………………………....33

Figure 12. Electivity values (Ei) for all genera found within this study..………………………...34

Figure 13. Electivity values (Ei) for genera found within five or more sites……………………..35

Figure 14. Regressions of L. maackii electivity on average canopy openness, beaver residency time (years) per site, and density of preferred stems among the eight sites……………………..36

List of Appendices Appendix 1. The total number of stems, stems cut, proportion cut, proportion of total stems cut and electivity (Ei) for each genus at each site…………………………………………………..37 Appendix 2. R code and output for binomial generalized regressions and linear regressions….41 Appendix 3. Barplot of the proportion of L. maackii stems cut within each size class for stems close to the water’s edge and stems far from the water’s edge………………………………….48

Acknowledgements

I would first like to thank Dr. David Gorchov for his invaluable insight and attentiveness throughout this entire project.

I would like to thank Zoey Armstrong, Zoey Scancarello, Kelly Benoit, Courtney Dvorsky, Anna Bowen, Nicole Berry, and Jake Godfrey for their assistance in the field, Mike Mahon for his help with statistics, and the Gorchov and Stevens labs for providing helpful feedback on study design and presentations.

I also thank Dr. Bartosz Grudzinski and Dr. Susan Hoffman for serving on my committee and providing insightful feedback as well as access to labs and field equipment.

I am grateful for financial support from the John L. Vankat Student Grant Fund for Field Research in Terrestrial Plant Ecology.

Lastly, I would like to thank the staff at Great Parks of Hamilton County, Five Rivers Metroparks, the Ohio Department of Natural Resources, the Oxford Department of Service and Engineering, and the Miami University Natural Areas for answering all my questions and giving me permission to conduct research at their sites.

vii

INTRODUCTION

Throughout the United States, riparian zones, interfaces between waterways and land, have become a conservation concern for land managers. Riparian zones have large ecological importance because they maintain the thermal regulation of water and establish microclimates, supply coarse organic matter to aquatic systems, provide large woody debris that affects the channel shape and water flow, provide refugia to fish and substrate for aquatic invertebrates, purify water and regulate nutrient inputs into streams and lakes, dampen floods and droughts by decreasing surface runoff and replenishing groundwater, and stabilize stream banks and stream- bed sediments (Naiman et al. 2005). The management goals for riparian areas increasingly include conservation of both terrestrial and aquatic wildlife because riparian zones have been found to support high levels of biodiversity on small areas of the landscape (Dudgeon et al. 2006, Stoffyn-Egli and Willison 2011).

One riparian obligate is the North American beaver, Castor canadensis. Beavers are considered keystone species within the riparian zone because they are ecosystem engineers that drastically alter hydrology, create and maintain riparian areas, and create ecosystem heterogeneity (Naiman et al. 1986). Beavers can alter the riparian zone by changing the water’s flow to create pools upstream of dams, which create new wetland areas (Naiman et al. 2005, Rosell et al. 2014). Because of this shift in water flow, there can be changes in the amount of erosion and thus sediments entering the water column (Rosell et al. 2014) and the vegetation type can change from terrestrial to aquatic or hydrophilic (Ray et al. 2001). Through selective foraging, beavers are also able to change the composition of riparian terrestrial vegetation (Martell et al. 2006).

Beaver Foraging Behavior

Beavers cut woody stems for use in both their dams and lodges, and as a food source. They have commensal microbiota in their digestive tracts and are capable of digesting and deriving energy from 30% of available cellulose (Currier et al. 1960). A study that enriched beaver gut microbiota in cultures indicated that the gut microbiota contained some enzymes that can breakdown lignocellulose (Wong et al. 2016). This aids beavers in digesting woody stems.

Beavers are central place foragers that become more selective the farther away they move from their dam (Jenkins 1980, Gallant et al 2004). They are also considered selective generalist herbivores because of their known preference for some species of woody plants over others (Jenkins 1975, Jenkins 1979, Gallant et al 2004). Beavers rarely forage farther than 50 meters from the water’s edge even if the colony has inhabited the site for a relatively long period of time (Stoffyn-Egli and Willison 2011). Seasonally, beavers increase wood cutting in late fall and early spring, when the ice melts and before the vegetation has leaves, to form food caches (Jenkins 1979, Brzyski and Schulte 2009). They have also been observed consuming more deciduous trees in the fall while preferring pines in the spring because of the greater seasonal nutrient concentration stability seen in coniferous trees (Jenkins 1979). The woody species most preferred by beaver in previous studies include Populus spp., Salix spp., Quercus rubra and Q. alba, Betula alleghaniesis and B. populifolia, Pinus spp., Hamamelis sp., Ostrya virginiana, and Carpinus caroliniana (Henry and Bookhout 1970, Jenkins 1979, Gerwing et al. 2013). 1

Beavers have been found to consume both aquatic and terrestrial herbaceous species. In a study conducted in Nebraska, aquatic herbs made up 55% of the beaver’s diet and were consumed year-round, but terrestrial vegetation was consumed in higher amounts in the summer (Severud et al. 2013). A study in a subarctic environment found that 60-80% of beavers diet was comprised of aquatic vegetation (Milligan and Humphries 2010). In southeast Ohio, Svendsen (1980) found that beavers fed on aquatic vegetation about 90% of the feeding time in the summer and 40-50% of the feeding time in early spring and fall. In that same study, beavers were observed feeding on Christmas fern (Polystichum acrostichoides) fronds and rhizomes extensively in January and February.

Herbivore/Plant Interactions

Through heavy grazing or browsing, herbivores can change the composition of palatable and unpalatable species in an area (Tierson et al. 1966, Marquis 1981). If an herbivore is selecting primarily for invasive plant species, this could lead to a prevention of that invasion, contributing to the biotic resistance hypothesis (BRH) (Elton 1958). In contrast if the herbivore is avoiding or has low preference for the invasive plant species, that species could succeed at invasion because of a lack of predators or pathogens, contributing to the enemy release hypothesis (ERH) (Maron and Vila 2001).

Beavers have been observed consuming and using invasive woody species in riparian areas in the southwest United States. A study conducted on the Rio Grande River in the Chihuahuan Desert found beavers had a high preference for the invasive Russian olive (Elaeagnus angustifolia) (Barela and Frey 2016). In areas with high cottonwood densities in eastern Montana, beaver cut 80% of the cottonwood trees while rarely cutting the invasive Russian olive or Salt Cedar (Tamarisk sp.), and growth rates of both Russian olive and tamarisk were higher in sites where beaver greatly reduced cottonwood canopy cover, suggesting that beaver were facilitating their ecological release (Lesica and Miles 2004). In Grand Canyon National Park, Mortenson et al (2008) found that in sites where beavers were abundant, tamarisk was also abundant, providing evidence that beavers facilitate tamarisk invasion.

In the eastern United States, only one study has examined beaver use of non-native woody plants. In the Appalachian Mountains of North Carolina, Rossell et al (2014) found that the nonnative shrub, Chinese privet (Ligustrum sinense), was one of the species ‘moderately selected’ by beaver, with a lower percentage cut than musclewood (Carpinus carolina), sweetgum (Liquidambar styraciflua), and tag alder (Alnus serrulata). I have not found any other papers reporting beaver use of invasive woody plants in the eastern hardwood forests. To my knowledge, there is no information on the interaction between beavers and Amur honeysuckle (Lonicera maackii), an invasive shrub common in Midwestern forests.

Amur honeysuckle, Lonicera maackii

Amur honeysuckle (Lonicera maackii) is a shrub from east Asia that was introduced in North America in 1898 as a horticultural species and for erosion control (Luken and Thieret 1996, McNeish and McEwan 2016). It is now regulated as an invasive species in eight states

(EDDMaps 2016). It expands leaves in early spring and loses leaves in late fall (McEwan et al. 2009, Wilfong et al. 2009), resulting in ‘extended leaf phenology’ (Smith 2013) compared to native woody plant species. Studies have shown that Amur honeysuckle reduces plant species richness and abundance in invaded ecosystems and can greatly reduce herbaceous plant fitness through above ground competition and allelopathy (literature reviewed by McNeish and McEwan 2016). In a study conducted in southwestern Ohio, L. maackii was found in lower densities in bottomlands compared to west and east facing slopes (Gayek and Quigley 2001). The authors hypothesized that these findings were due to more continuous canopy cover in bottomlands (Gayek and Quigley 2001). Lonicera maackii is also positively associated with riparian urbanization (Pennington et al 2010, White et al 2014).

Beaver Abundance in Ohio

Beavers were abundant in Ohio until their extermination in 1830 (Chapman 1949). They disappeared from the Cincinnati area by 1805 (Hedeen 1985). Beavers re-established in parts of northwestern and eastern Ohio in the early 1930’s, immigrating from Michigan and Pennsylvania (Chapman 1949) (Figure 1). Possible introductions from sportmens’ organizations and escape from fur farms also aided beavers in colonizing parts of Ohio (Chapman 1949). Beavers reappeared in the Cincinnati area in 1984 (Hedeen 1985) and the overall Ohio beaver population has been growing since the 1980s (Figures 2, 3).

In December 2017, I observed a beaver dam that was built on Collins Run in Peffer Park of the Miami University Natural Areas. I found many stems of L. maackii shrubs as well as saplings of other woody species cut by beaver. I could see L. maackii intertwined with other woody species within the dam infrastructure. Other graduate students have also observed cut stems of L. maackii around Bachelor Pond.

Potential importance of beaver-Lonicera interactions in riparian zone

Alterations in the riparian plant community due to L. maackii invasion can have impacts on both the terrestrial and aquatic environment. Lonicera maackii has been found to negatively correlate with native tree seedlings and sapling densities within urban or disturbed riparian zones (White et al. 2014). A study conducted by McNeish et al (2015) in western Ohio found that where L. maackii has been removed from a stream reach, there is significantly greater in-stream light availability and terrestrial organic matter input into riparian streams. Beavers could be facilitating L. maackii invasion by cutting the native woody competitors (as found in some studies with other invasive species (Lesica and Miles 2004, Mortenson et al. 2008)) or could hinder the invasion by removing L. maackii and allowing native plant species to grow.

Study Questions

I investigated whether North American beavers use L. maackii preferentially compared to other woody species and what accounts for differences in preference for this invasive among sites. To test these questions, I surveyed cut, browsed, and uncut woody stems along transects at eight sites in southwest Ohio and measured the importance of average canopy openness, beaver

3 residency time, and the density of preferred stems in explaining differences in preference for L. maackii among sites.

METHODS

Eight sites in southwest Ohio (Figure 4, Table 1) were selected for this study. Sites were located in the southern tip of the beech-maple-basswood region of eastern deciduous forest (Dyer 2006) in secondary forest stands within a matrix of row-crop agriculture and development. Lonicera maackii is a very common understory shrub in these forest stands. The predominant soil series within this region of Ohio is Bennington-Cardington-Centerburg, which are fine textured soils formed in glacial till with low lime content and range from poorly to moderately well drained (ODNR n.d.). The Climate Normals (1981-2010) for a point close to the center of the study area, Eaton, Ohio, are an annual temperature of 10.7˚C and annual precipitation of 1042 mm (U.S. Climate Data 2019).

All sites were spaced at least 1 km apart, following methods in Small et al (2016), in order to avoid sampling the same beaver colony twice. For a site to be suitable for this study it had to be forested, have woody stems recently cut by beaver, and have L. maackii present.

At each site, I sampled woody stems in fall (15 October- 30 November, 2018) and spring (25-31 March, 2019) using modifications of the transect method used by Barela and Frey (2016), Gerwing et al (2013), and Gallant et al (2004). Only woody stems that were greater than 2.5 cm in diameter were considered (Jenkins 1975). At each site I marked off either one (240 m) or two (120 m) primary transects and six secondary transects (each 25 m); all transects were 2 m wide.

Transect Configuration

The configuration of transects at a site depended on the waterway and forest characteristics at each site. If the site was a stream with a forested riparian zone on both banks that was > 25 m wide, I had two 120 m primary transects that ran parallel to the water’s edge and were located 0.5 m into the forest vegetation (Figure 5A). The primary transects were centered in the area of highest woody stem cutting so that 60 m of the transect was upstream and 60 m of the transect was downstream of the center of cut woody stems. I then had six, 25 m secondary transects perpendicular to the primary transects. The secondary transects were located at the center of the primary transect and 40 m upstream and downstream along the primary transect. If the site was a stream with a forested riparian zone on one bank that was > 25 m wide while the other bank had a narrow tree/shrub riparian zone < 25 m wide, I had two 120 m primary transects that ran parallel to the water’s edge, 0.5 m into the beginning of the riparian vegetation (Figure 5B). The primary transects were centered in the area of highest woody stem cutting so that 60 m of the transect was upstream and 60 m was downstream. I then had six, 25 m secondary transects perpendicular to the primary transect on the side with the forested riparian zone > 25 m. If the site was a pond or lake, I had one 240 m primary transect parallel to the water’s edge, 0.5 m into the beginning of the riparian vegetation (Figure 5C). The primary transect was centered in the area of highest woody stem cutting. The six 25 m secondary transects were perpendicular to the primary transect.

Woody Plant Sampling

During the fall 2018 woody vegetation survey, all woody stems > 2.5 cm in diameter were identified, measured for diameter at 30 cm or, if < 30 cm, at the highest height possible following methods found in Gallant et al. (2004), and scored as cut, browsed, or uncut. To be considered “cut”, the stem had to be cut completely through by beavers, and for it to be considered “browsed”, the stem had cut marks made by beavers but not cut all the way through. The wood of the cut stem also had to be light tan to gray-brown in color, to exclude stems that had been cut well before (> ~ 1 year) the census. I measured the distance the stem was from the water’s edge and the distance along the transect to the nearest 0.1 m.

In spring 2019, I resurveyed the stems that were not cut in the fall survey, and scored these as cut, browsed, or uncut. Because there were so few browsed stems, I included these stems in the uncut stem counts.

Species were identified with field guides and cut woody stems were identified using a bark photo collection I created from uncut stems. I distinguished between native and non-native taxa in my dataset using the U.S. Department of Agriculture’s PLANTS database, plants.sc.egov.usda.gov.

For data analysis, I grouped all species by genus because in some genera, species identification was difficult and in others, individual species were uncommon. Acer negundo was analyzed separately from other Acer species because this species was abundant and was cut at a higher proportion compared to other Acer stems.

Canopy Openness

I measured the canopy openness at each site during August 2018 using a spherical densitometer Model C (Strickler 1959), utilizing methods found in Lemmon (1956). Spherical densitometers are comparable to more sophisticated measures such as hemispherical photography in their effectiveness at measuring light environments in forested ecosystems (Englund et al. 2000). I took these measurements at 1.2 m (chest height) at the middle and each end of each secondary transect, for a total of 18 samples at each site. Values were averaged to obtain an average canopy openness value for each site.

Beaver Residency Time

In order to obtain beaver residency times at each site, I contacted park managers and naturalists and requested information on how long beavers had been cutting at that site. The dates I obtained were converted to number of years beavers were present at that location as of October 2018.

Density of Preferred Stems

Stems in genera that were consistently preferred by beaver (had an electivity (see below) value > 0 at every site they were present) and with a diameter between 25-69 mm were classified

5 as “preferred stems.” For each site, the number of these stems divided by the total sampled area (780 m2) resulted in the density of preferred stems.

Photo documentation of beaver cutting

Browning Strike Force Elite HD (Model BTC-5HDE) trail cameras were placed at some sites in August 2018-mid September 2018, and pointed near areas of heavy beaver cutting in order to document beaver cutting of L. maackii and other woody species (Figure 6).

STATISTICAL ANALYSIS

Probability of a Stem being Cut

To determine whether beavers cut woody stems based on genus or other factors, I conducted binomial generalized regressions (logistic regressions) to determine which factors determine the probability of a woody stem being cut (Gerwing et al. 2013). Included as fixed factors were the distance the woody stem was from the water’s edge, the diameter of the woody stem, the genus identity, and the interaction of diameter and distance, while site was included as a random factor. I conducted three separate binomial generalized regressions throughout this study using R 3.4.3 (R Core Team 2017), the lme4 package (Bates et al. 2015), and used the car package (Fox and Weisberg 2011) to conduct an ANOVA on each regression. The first regression used all stems and the second used only stems of genera that appeared at five or more sites, in order to exclude genera present only at a few sites that could potentially skew results from the first regression. The third regression used only L. maackii stems and genus was not included as a factor.

Electivity

In order to determine which genus beavers were choosing to cut, I used an electivity index for each woody genus available at each site during the fall sampling period. Electivity is a measure of preference based on the proportion that is consumed of a food item compared to that food item’s abundance in the community. To calculate electivity, I used the formula (Equations 1 and 2) from Vanderploeg and Scavia (1979):

1) 1 1 Ei= (푊 − ) ൗ(푊 + ) 푖 푛 푖 푛

In these formulas, ri is the proportion of stems cut that belong to genus i, pi is the proportion of stems available that belong to genus i, and n is the number of genera available at that site. Wi is the selectivity coefficient. If the value of electivity (Ei) is between 0 and 1, then that genus is preferably cut by beaver and if the value is between -1 and 0, beavers are avoiding 6 that genus. A value of 0 indicates that beavers are neither avoiding nor preferentially cutting that genus.

Linear Regressions

In order to determine which environmental factors or site characteristics were influencing L. maackii electivity among the 8 sites, I conducted linear regressions of L. maackii electivity on each of these factors: average canopy openness, residency time of the beaver, and density of preferred stems. All data were analyzed using R 3.4.3 (R Core Team 2017).

RESULTS

A total of 2,344 stems comprised of 32 genera were observed over the eight study sites in Fall 2018 (Appendix 1). Of these, 1,837 (78%) were uncut and re-censused in Spring 2019.

Factors Influencing Stem Cutting

Fall 2018

For all stems in the dataset, genus, distance to the water’s edge, and stem diameter were significant predictors in determining whether a stem was cut or uncut (Table 2, Appendix 2). The interaction between distance to the water’s edge and stem diameter was not significant (Table 2, Appendix 2). The probability of beavers cutting stems was highest near the water’s edge and decreased the farther away the stem was from the water’s edge (Figure 7). There was also higher probability of a stem being cut if its basal diameter was within the 25-69 mm size class, and the probability decreased with increasing diameter size classes (Figure 8).

For the genera found at five or more sites, genus, distance to the water’s edge, and stem diameter were significant predictors in determining whether a stem was cut or uncut (Table 3, Appendix 2) as seen in the previous model which included all genera. As in the previous model, the interaction between distance from the water’s edge and stem diameter was not significant (Table 3, Appendix 2).

For the analysis of only L. maackii stems, genus, distance to the water’s edge, and stem diameter were significant predictors in determining whether a stem was cut or uncut (Table 4, Appendix 2) as was seen in the models that included other genera. The probability of a stem getting cut was higher closer to the water’s edge and decreased the farther that stem was from the water’s edge (Figure 9). As seen in the previous models, L. maackii stems had a higher probability of being cut if they were in the 25-69 mm size class (Figure 10). The interaction between distance to the water’s edge and stem diameter was significant (Table 4, Appendix 2). The negative relationship between the proportion of stems cut and stem diameter was stronger for stems closer to the water’s edge (Appendix 3).

Only 41 stems were browsed in the fall census. These consisted of 13 genera of which Acer, Fraxinus, and Lonicera had the most browsed stems (Table 5). The diameter of these stems ranged from 27 mm – 1197 mm, with most being > 70 mm. 7

Spring 2019

Only 14 stems were cut and 9 browsed in the Spring 2019 census of stems not cut or browsed in Fall 2018 (Table 6). More than half of these stems were between 25-39 mm in diameter and half were from the genus Lonicera (Table 6).

Electivities

Lonicera maackii accounted for 41% of stems cut in Fall 2018, followed by Elaeagnus which accounted for ~20% of cut stems, and A. negundo which accounted for ~10% of cut stems (Figure 11). Despite L. maackii accounting for many cut stems, electivities revealed it was not preferred at six of the eight sites (Appendix 1). Lonicera maackii electivity ranged from complete avoidance (-1) at Hueston North to slight preference (0.11) at Miami Whitewater Forest and Acton Lake.

The genera Salix, Ligustrum, Pyrus, Carya and the species Acer negundo were consistently preferred (Ei > 0) at all sites in which they were found (Figure 12), and they comprise the genera for preferred stems. Additionally, the genera Fraxinus and Elaeagnus, had mean electivities (among sites) > 0 (Figures 12 and 13).

Differences in L. maackii Electivity Among Sites

There was a slight positive relationship between L. maackii electivity and average canopy openness, but it was not significant (p = 0.158, R2 = 0.1861) (Figure 14a). Lonicera maackii electivity had a negative relationship with beaver residency time in years, but this was not significant (p = 0.1559, R2 = 0.1889) (Figure 14b). Electivity of L. maackii also had a negative relationship with density of preferred stems and this interaction was marginally significant (p = 0.071, R2 = 0.3514) (Figure 14c).

DISCUSSION

Though L. maackii made up the largest proportion of cut stems in the fall census, it was not preferred by beaver at most sites. Consistency among sites in electivity values for a genus increases the confidence that that genus is really preferred or avoided. So because beavers are not cutting L. maackii preferentially compared to other native woody species at most sites, there is a high level of confidence that it is predominantly avoided by beavers and there is no evidence that beavers are hindering L. maackii invasion. Another invasive species found within my transects, Autumn olive (Elaeagnus umbellata), was preferentially cut by beavers at 5 of 7 sites where it was present, so there is a high level of confidence that beaver are preferentially cutting this genus. This finding suggests that beavers may be hindering the invasion of E. umbellata in riparian areas. Beaver have been shown to have a preference for another species in this genus, Russian Olive (Elaeagnus angustifolia) (Barela and Frey 2016). Because of findings from this study, where beaver preferentially cut E. umbellata, and the previous study, where beavers preferentially cut E. angustifolia, there is evidence that beaver hinder the invasion of both Elaeagnus species. Another invasive species, border privet (Ligustrum obtusifolium), was preferentially cut at the one study site where it was found, providing some evidence beaver 8 potentially hinder L. obtusifolium invasion. A previous study found beavers had a moderate selection for Chinese privet (Ligustrum sinense) (Rossell et al 2014). Further investigation of more sites where Ligustrum is present is needed to test for evidence of beavers hindering Ligustrum invasion.

My observations of beaver-cut L. maackii stems lead me to infer that beavers are using L. maackii primarily as building material in their lodges and dams rather than a food source. In spring 2018 when I was first scouting sites, I observed cut L. maackii branches that appeared to have been dragged towards the water’s edge. These stems still had leaves and none of the bark had been stripped off. I also noticed that both the dam at Peffer Park and lodge at Bachelor Pond contained many L. maackii stems of various sizes. A study conducted by Raffel et al. (2009) found that beavers selected (overall – for building material and food) for medium-sized stems (2.0-6.9 cm) at all distances from the water’s edge, which aligns with beaver cutting of L. maackii in this study. As central place foragers, beaver should be more selective for size classes the farther they are foraging from their central place. However, I found that for L. maackii stems, beavers were more selective for small size classes closer to the water’s edge (Appendix 3) which is the opposite of patterns found by Gallant et al. (2004) and Raffel et al. (2009). Because beavers are not following the pattern of a central place forager when cutting L. maackii at my study sites, this is indicative that they are using L. maackii primarily for use as building material as opposed to food.

Native taxa such as Acer negundo, Salix, Fraxinus, and Carya had mean electivities that were above zero, indicating that they were predominantly preferred at most sites. Acer and Salix have been found to be preferred by beaver in other sites (Henry and Bookhout 1970, Jenkins 1979, Gerwing et al. 2013), while Fraxinus was found to be neutrally preferred, and Carya, depending on the species, was either avoided or neutrally preferred (Raffel et al. 2009).

Many factors could be influencing beaver preference for certain genera over others, such as differences in production of secondary compounds, particularly tannins, and in nutrition obtained by cutting certain stems. Tannins can bind to plant proteins and inhibit herbivore digestive enzymes, making it harder for the herbivore to digest plant material (Zucker 1983), which has been observed in white-tailed deer (Odocoileus virginianus), moose (Alces alces), and elk (Cervus canadensis) (Robbins et al. 1987). Beavers have been found to produce proteins that can bind to linear condensed tannins found in highly preferred foods, such as Populus and Salix species, but cannot bind to other tannins such as branched condensed and hydrolysable tannins (Hagerman and Robbins 1993). However, when condensed tannins are in high concentrations in Populus species, beavers were found to forage on these species less (Bailey et al. 2004). Other generalist rodent herbivores such as eastern gray squirrels (Sciurus carolinensis) have been found to show a preference for low tannin food like Quercus alba compared to high tannin foods like Juglans nigra and Quercus rubra (Barthelmess 2001).

Beaver could be cutting certain stems based on differences in nutritional quality and digestibility among genera. Doucet and Fryxell (1993) found that in a cafeteria experiment where they were presented with five species, beavers chose an energetically optimal diet by consuming more nutritious and easily digested trembling aspen (Populus tremuloides), white water lily (Nymphaea odorata), and raspberry (Rubus idaeus), while only sparsely choosing speckled alder

(Alnus rugosa) and red maple (Acer rubrum), because the latter species are harder to digest and offer fewer nutrients. Another study by Fryxell and Doucet (1993) found that beavers consumed low nutrient and hard to digest A. rubrum stems only after preferred high nutrient and easily digested stems of P. tremuloides and A. rugosa stems were depleted.

Another factor that could be influencing beavers stem selection could be the energy costs and benefits of cutting certain stems. Beavers have also been shown to cut more preferred species relative to low preference species the farther they travel from their lodge (Raffel et al 2009). This cutting behavior has been shown in many studies (McGinley and Whitham 1985, Gallant et al. 2004, Martell and Cumming 2006, Stoffyn-Elgi and Willison 2011, Salandre et al. 2017) and is often attributed to beavers being central place foragers. As central place foragers, beavers are expected to cut fewer stems and be more selective the farther they are from the water’s edge in order to decrease energy costs and increase nutritional benefits of obtaining stems (McGinley and Whitham 1985). In this study, the probability of a stem being cut, in both the all stem and L. maackii stem analyses, was higher closer to the water’s edge.

Difference among sites:

Differences in L. maackii electivity at each site could be explained by the density of preferred stems. Because of selectivity by beaver, if there is a higher density of preferred stems at a site, beavers are more likely to eat the preferred stems and avoid L. maackii stems. However, if the density of preferred stems is low at a site, L. maackii is either preferred by beaver or neutrally selected by beaver. Beavers at my study sites could be selecting first for more preferred genera that have better nutrient quality and that are easier to digest before choosing less preferred genera that are not as nutritious or digestible.

There was not a significant relationship between L. maackii electivity and average canopy openness at each site, which could be related to the limited range of canopy openness values among sites. Most sites had low canopy openness values (between 0-10%) and none of my sites had a canopy openness value greater than 40%. Lonicera maackii electivity also did not have a significant relationship with beaver residency time. Sites that had longer beaver residency times had L. maackii electivities that ranged from complete avoidance (-1) to preference (0.11). Though beavers differed in their colonization time, this factor does not necessarily determine which genus they are choosing to cut, but it could influence the number of stems they cut at a site.

Few cut Stems in Spring Census:

The very small number of stems found cut or browsed in the spring census may have been due beavers shifting activity away from my transects. When I originally selected spots to lay my transects, I centered my primary transects on areas that had the most cutting at the site. The beavers could have exhausted most of the available preferred stems along my transects before the fall census and moved on to other sections of the site. This would be relatively easy for the beaver to do because most of my sites connect to or are adjacent to waterways in the area. When I was conducting the spring stem census, I noticed freshly cut stems upstream and

10 downstream of my original transects at most of my sites. This leads me to believe that the beavers may have simply moved downstream or upstream of my sites or to a new location.

Another reason why beavers cut so few stems at my sites could be that the beavers are still utilizing their cached resources and therefore did not need to cut new stems just yet. A study conducted in southeastern Ohio found that herbaceous vegetation made up between 40-50% of the beaver’s diet in early spring and that material in caches and bark from previously downed trees along the shoreline made up a large portion of the beaver’s winter food (Svendsen 1980). When I sampled this spring, there was still very little herbaceous vegetation, so the beavers could still be using their winter caches.

Future Research:

During the fall census, on a few of the beaver-cut L. maackii stems, I observed deer browse on some of the shoots of new growth. Lonicera maackii shoots have been observed to appear at three weeks after the stem is cut, and they grow three-fold within a year of the stem being cut (McDonnell et al. 2005). If the beavers at my sites cut L. maackii in early spring during one of their peak cutting seasons (Jenkins 1979, Brzyski and Schulte 2009), by summer the shoot should be big enough for other herbivores such as white-tailed deer to browse the new stem growth. White-tailed deer have been shown to browse L. maackii, particularly in early spring when other preferred food sources are not abundant and in late summer (Martinod and Gorchov 2017), right before I made these observations. In my study sites, the combined cutting of beaver and browse by deer could potentially hinder overall L. maackii plant growth. This interaction between beavers and another herbivore has also been observed with beaver and elk and their combined effects on willows. Baker et al (2005) found that willows at sites that had combined herbivory by beaver and elk had a smaller biomass and diameter, were shorter, and had a higher percentage of dead biomass that strongly suppressed standing crop, compared to sites that only had one or no herbivore present. This interaction between beaver and white-tailed deer at my study sites warrants further investigation into the possibilities of it serving as a natural management of L. maackii. It is also possible that beaver cutting of L. maackii stems creates a potentially important food source (new shoots) for white-tailed deer.

Another topic I would like to address in future studies is management techniques that could promote beaver cutting of L. maackii stems. In a study exploring promotion of beaver cutting of invasive tamarisk (Tamarix ramosissima), Kimball and Perry (2008) found that when casein hydrolysate, an anti-herbivory compound, was applied to cuttings of two highly preferred species, black poplar (Populus nigra) and Scouler’s willow (Salix scouleriana), and these cuttings were presented with unpreferred tamarisk, beavers increased consumption of tamarisk cuttings. Further investigation into whether applying casein hydrolysate to native stems at my sites will increase beaver cutting of L.maackii stems could determine whether this is an effective management technique.

CONCLUSIONS

Although L. maackii stems accounted for 41% of the total stems that beavers cut, it was low preference at 6 out of 8 of my study sites. Because of these findings, there is no evidence 11 that beavers are hindering L. maackii invasion. Another invasive species, Autumn olive (Elaeagnus umbellata), was preferred at 5 of the 6 sites where it was found and a third, border privet (Ligustrum obtusifolium) was preferred at the one site where it was found, which suggests that beavers hinder the invasion of these two species, at least at some sites. The relationship between L. maackii electivity at each site and the density of “preferred” stems (consistently preferred genera between 25-69 mm in diameter) was marginally significant. Beavers at sites with a higher density of preferred stems are predominantly cutting those stems and avoiding L. maackii stems, while beavers at sites with a low density of preferred stems are preferentially cutting or neutrally preferring to cut L. maackii stems.

REFERENCES

Bailey JK, Schweitzer JA, Rehill BJ, Lindroth RL, Martinsen GD, Whitham TG. 2004. Beavers as molecular geneticists: a genetic basis to the foraging of an ecosystem engineer. Ecology 85:603-608. Baker BW, Ducharme HC, Mitchell DCS, Stanley TR, Peinetti HR. 2005. Interaction of beaver and elk herbivory reduces standing crop of willow. Ecological Applications 15:110- 118. Barela IA, Frey JK. 2016. Habitat and forage selection by the American beaver (Castor canadensis) on a regulated river in the Chihuahuan Desert. The Southwestern Naturalist 61:286-293. Barthelmess EL. 2001. The effects of tannin and protein on food preference in eastern grey squirrels. Ethology Ecology & Evolution 13:115-132. Bates D, Maechler M, Bolker B, Walker S. 2015. Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67:1-48. Brzyski JR, and Schulte BA. 2009. Beaver (Castor canadensis) impacts on herbaceous and woody vegetation in southeastern Georgia. The American Midland Naturalist 162: 74- 86. Chapman, FB. (1949). The beaver in Ohio. Journal of Mammalogy 30:174-179. Currier A, Kitts WD, Cowan IMcT. 1960. Cellulose digestion in the beaver (Castor canadensis). Canadian Journal of Zoology 38:1109-1116. Doucet CM, Fryxell JM. 1993. The effect of nutritional quality on forage preference by beavers. Oikos 67:201-208. Dudgeon D, Arthington AH, Gessner MO, Kawabata Z, Knowler DJ, Leveque C, Naiman RJ, Prieur-Richard A, Soto D, Stiassny MLJ, Sullivan CA. 2006. Freshwater biodiversity: importance, threats, status, and conservation challenges. Biological Reviews 81:163-182. Dyer JM. 2006. Revisiting the deciduous forests of Eastern North America. BioScience 56:341- 352. EDDMapS. 2017. Amur honeysuckle (Lonicera maackii). Early Detection and Distribution Mapping System. The University of Georgia – Center for Invasive Species and Ecosystem Health. http://www.eddmaps.org/Species/subject.cfm?sub=3040#list. Accessed 31 Jan 2018. Elton CS. 1958. The ecology of invasions by plants and animals. Methuen. Englund SR, O’Brien JJ, Clark DB. 2000. Evaluation of digital and film hemispherical photography and spherical densiometry measuring forest light environments. Canadian Journal of Forest Research 30:1999-2005. Fox J, Weisberg S. 2011. An {R} companion to applied regression. Second Edition. Thousand Oaks CA: Sage. URL: http://socserv.socsci.mcmaster.ca/jfox/Books/Companion. Fryxell JM, Doucet CM. 1993. Diet choice and the functional response of beavers. Ecology 74:1297-1306. Gallant D, Berube CH, Tremblay E, Vasseur L. 2004. An extensive study of the foraging ecology of beavers (Castor canadensis) in relation to habitat quality. Canadian Journal of Zoology 82:922-933. Gayek A, Quigley MF. 2001. Does topography affect the colonization of Lonicera maackii and Ligustrum vulgare in a forested glen in Southwestern Ohio? The Ohio Journal of Science. 101:95-100. 13

Gerwing TG, Johnson CJ, Alstrom-Rapaport C. 2013. Factors influencing forage selection by the North American beaver (Castor canadensis). Mammalian Biology 78:79-86. Hagerman AE, Robbins CT. 1993. Specificity of tannin-binging salivary proteins relative to diet selection by mammals. Canadian Journal of Zoology 71:628-633. Hedeen, SE. (1985). Brief Note: Return of the Beaver, Castor Canadensis, to the Cincinnati Region. Henkin MA, Medley KE, Abbitt RJ. 2013. Invasion dynamics of nonnative Amur honeysuckle over 18 years in a southwestern Ohio forest. The American Midland Naturalist 170:335- 347. Henry DB, Bookhout TA. 1970. Utilization of woody plants by beavers in Northeastern Ohio. Ohio Journal of Science 70:123-127. Jenkins SH. 1975. Food selection by beavers: a multidimensional contingency table analysis. Oecologia. 21:157-173. Jenkins SH. 1979. Seasonal and year-to-year differences in food selection by beavers. Oecologia 44:112-116. Jenkins SH. 1980. A size-distance relation in food selection by beavers. Ecology 61: 740-746. Kimball BA, Perry KR. 2008. Manipulating beaver (Castor canadensis) feeding responses to invasive tamarisk (Tamarix spp.). Journal of Chemical Ecology 34:1050-1056. Lemmon PE. 1956. A spherical denisiometer for estimating forest overstory density. Forest Science 2: 314-320. Lesica P, Miles S. 2004. Beavers indirectly enhance the growth of Russian olive and tamarisk along eastern Montana rivers. Western North American Naturalist 64:93-100. Luken JO, Thieret JW. 1996. Amur honeysuckle, its fall from grace: lessons from the introduction and spread of a shrub species may guide future plant introductions. BioScience 46: 18-24. Maron JL, Vila M. 2001. When do herbivores affect plant invasion? Evidence for the natural enemies and biotic resistance hypothesis. Oikos 95:361-373. Marquis DA. 1981. Effect of deer browsing on timber production in Allegheny hardwood forests of northwestern Pennsylvania. U.S. Forest Service Research Paper NE-465. Martell, KA, Foote, AL, Cumming SG. 2006. Riparian disturbance due to beavers (Castor canadensis) in Alberta’s boreal mixedwood forests: implications for forest management. Ecoscience 13: 164-171. Martinod KL, Gorchov DL. 2017. White-tailed deer browse on an invasive shrub with extended leaf phenology meets assumptions of an apparent competition hypothesis. AoB Plants 9:1-14. McDonnell AL, Mounteer AM, Owen HR, Todd BL. 2005. Influence of stem cutting and glyphosate treatment of Lonicera maackii, and exotic and invasive species, on stem regrowth and native species richness. Transactions of the Illinois State Academy of Science 98:1-17. McEwan RW, Birchfield MK, Schoergendorfer A, Arthur MA. 2009. Leaf phenology and freeze tolerance of the invasive shrub Amur honeysuckle and potential native competitors. The Journal of the Torrey Botanical Society 136:212-221. McGinley MA, Whitham TG. 1985. Central place foraging by beavers (Castor canadensis): a test of foraging predictions and the impact of selective feeding on the growth form of cottonwoods (Populus fremontii). Oecologia 66:558-562.

McNeish RE, McEwan RW. 2016. A review on the invasion ecology of Amur honeysuckle (Lonicera maackii, Caprifoliaceae) a case study of ecological impacts at multiple scales. Journal of the Torrey Botanical Society 143: 367-385. McNeish RE, Moore EM, Benbow ME, McEwan RW. 2015. Removal of the invasive shrub, Lonicera maackii, from riparian forests influences headwater stream biota and ecosystem function. River Research and Applications 31:1131-1139. Milligan HE, Humphries MM. 2010. The importance of aquatic vegetation in beaver diets and the seasonal and habitat specificity of aquatic-terrestrial ecosystem linkages in a subarctic environment. Oikos 119: 1877-1886. Mortenson SG, Weisburg PJ, Ralston BE. 2008. Do beavers promote the invasion of non-native Tamarix in the Grand Canyon riparian zone? Wetlands 28:666-675. Naiman R, Melillo J, Hobbie J. 1986. Ecosystem alteration of boreal forest streams by beaver (Castor canadensis). Ecology 67:1254-1269. Naiman RJ, Decamps H, McClain ME. 2005. Riparia: Ecology, conservation, and management of streamside communities. Elsevier Academic Press, Boston, MA. 430 p. ODNR. n.d. Soil regions of Ohio. Division of Soil and Water Conservation. https://water. ohiodnr.gov/portals/soilwater /pdf/soil/Soil_Regions of_Ohio brochure.pdf. Accessed 10 July 2019. Pennington DN, Hansel JR, Gorchov DL.2010. Urbanization and riparian forest woody communities: diversity, composition, and structure within a metropolitan landscape. Biological Conservation 143:182-194. Raffel TR, Smith N, Cortright C, Gatz AJ. 2009. Central place foraging by beavers (Castor canadensis) in a complex lake habitat. The American Midland Naturalist 162:62-74. Ray AM, Rebertus AJ, Ray HL. 2001. Macrophyte succession in Minnesota beaver ponds. Canadian Journal of Botany 79:487-499. R Core Team. 2017. R: A language and environment for statistical computing. R foundation for statistical computing. Vienna, Australia. URL http://www.R-project.org/. Robbins CT, Hanley TA, Hagerman AE, Hjeljord O, Baker DI, Schwartz CC, et al. 1987. Role of tannins in defending plants against ruminants: reduction in protein availability. Ecology 68:98-107. Rossell Jr. CR, Arico S, Clarke HD, Horton JL, Ward JR, Patch SC. 2014. Forage selection of native and nonnative woody plants by beaver in a rare-shrub community in the Appalachian Mountains of North Carolina. Southeastern Naturalist 13:649-662. Salandre JA, Beil R, Loehr JA, Sundell J. 2017. Foraging decisions of North American beaver (Castor canadensis) are shaped by energy constraints and predation risk. Mammal Research 62:229-239. Severud WJ, Belant JL, Windels SK. 2013. Seasonal variation in assimilated diets of American beavers. The American Midland Naturalist 169:30-42. Small BA, Frey JK, Gard CC. 2016. Livestock grazing limits beaver restoration in northern New Mexico. Restoration Ecology 24: 646-655. Smith LM. 2013. Extended leaf phenology in deciduous forest invaders: mechanisms of impact on native communities. Journal of Vegetation Science 24: 979-987. Stoffyn-Egli P, Willison JHM. 2011. Including wildlife habitat in the definition of riparian areas: the beaver (Castor canadensis) as an umbrella species for riparian obligate animals. Environmental Review 19:479-493.

Strickler GS. 1959. Use of the densiometer to estimate density of forest canopy on permanent sample plots. Forest Science 2:314-320. Svendsen GE. 1980. Seasonal change in feeding patterns of beaver in southeastern Ohio. The Journal of Wildlife Management. 44:285-290. Tierson WC, Patrick EF, Behrend DF. 1966. Influence of white-tailed deer on northern hardwood forest. Journal of Forestry 64:801-805. U.S Climate Data. 2019. Your Weather Service version 2.3. https://usclimatedata.com/climate/ eaton/ohio/united-states/usoh0284. Accessed 11 July 2019. Vanderploeg HA, Scavia D. 1979. Two electivity indices for feeding with special reference to zooplankton grazing. Journal of the Fisheries Board of Canada 36:362-365. White RJ, Carreiro MM, Zipperer WC. 2014. Woody plant communities along urban, suburban, and rural streams in Louisville Kentucky, USA. Urban Ecosystems 17:1061-1094. Wilfong BN, Gorchov DL, Henry MC. 2009. Detecting and invasive shrub in deciduous forest understories using remote sensing. Weed Science 57:512-520. Wong MT, Wang W, Lcourt M, Couturier M, Edwards EA, Master ER. 2016. Substrate-driven convergence of the microbial community in lignocellulose- amended enrichments of cut microflora from the Canadian beaver (Castor canadensis) and North American moose (Alces americanus). Frontiers in Microbiology 7:961. Zucker WV. 1983. Tannins: does structure determine function? An ecological perspective. American Naturalist 121:335-365.

Table 1: The location, city, beaver residency time, management entity, aquatic system, and transect type (Fig. 5), and field notes from spring/summer 2018 at each site.

Beaver Residency Transect Location City Mgt Entity System Notes Time Type (Years) HS cutting; beavers active Peffer Park Miami over winter and spring, Oxford Stream A (PEPK) 0.25 Natural Areas old dam washed out in spring Ohio Cutting from late spring, Hueston Hueston Department dam washed out in early North Stream A Woods 20 of Natural spring, still some fresh (HUNO) Resources activity Ohio Fresh cutting along bank Acton Lake Hueston 15 Department Lake B by dam in cove along the (ACLA) Woods of Natural west shore trail Resources Bachelor 3 beavers, lots of activity, Miami Pond Oxford 5 Pond B lodge along southern Natural Areas (BAWO) bank

Oxford Cutting around Memorial

Oxford Department Day, found fresh scat, 2.75 Landfill Oxford of Service Stream C utilize retention pond and

(OXLA) and Collins run behind

Engineering retention pond.

Miami Dam and den are built Great Parks Whitewater into stream; has pond and Harrison 5 of Hamilton Stream A Forest channels, lots of cutting County (MIWI) both native and hs

Possum Northern area of pond has Creek Five Rivers Dayton Pond B cut stems, regular beaver MetroPark 16 MetroParks activity in area, old lodge (POCR) Wegerzyn Built lodge in pond, Gardens Five Rivers Beaver Dayton A created channels, very MetroPark 4 MetroParks Channel recent activity (WEGA)

Table 2: Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut, using stems from all genera.

Chisq DF P Intercept 20.0208 1 <0.001 Genus 291.3942 24 <0.001 Distance to Water’s Edge 11.7042 1 <0.001 Diameter 46.7633 1 <0.001 Distance to Water:Diameter 0.0007 1 0.979

Table 3: Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut using stems from genera found in five or more sites.

Chisq DF P Intercept 23.3582 1 <0.001

Genus 276.4757 13 <0.001 Distance to Water’s Edge 11.2990 1 <0.001 Diameter 41.9363 1 <0.001 Distance to Water:Diameter 0.0014 1 0.9696

Table 4: Analysis of Deviance table (Type III Wald chisquare tests) for generalized binomial regression on whether a stem was cut or uncut using stems from L. maackii.

Chisq DF P Intercept 0.8711 1 0.351 Distance to Water’s Edge 24.3986 1 <0.001 Diameter 29.6135 1 <0.001 Distance to Water:Diameter 10.8096 1 0.001

Table 5: The number of browsed stems for each genus and each diameter class at each site during the fall census.

Diameter Class Genus 25-39.9mm 40-69.9mm >70mm Total Acer 0 0 13 13 Celtis 0 1 0 1 Crataegus 0 1 0 1 Fraxinus 0 0 6 6 Juglans 0 0 1 1 Juniperus 0 2 1 3 Lonicera 2 2 0 4 Platanus 0 0 3 3 Populus 0 0 1 1 Prunus 0 1 0 1 Quercus 1 0 2 3 Ulmus 0 0 3 3 Vitis 0 1 0 1 Total 3 8 30 41

Table 6: The number of cut and browsed stems during the spring re-census in each genus and each diameter class.

Diameter Class Genus 25-39mm 40-69mm >70mm Total Celtis 0 1 0 1

Elaeagnus 3 0 0 3 Juglans 0 0 1 1 Juniperus 0 1 2 3 Lonicera 8 3 0 11 Platanus 1 0 0 1 Quercus 0 0 1 1 Ulmus 0 0 1 1 Vitis 1 0 0 1 Total 13 5 5 23

Figure 1: First survey of beaver in Ohio 1949 after extermination in 1830. The author inferred beaver colonies in the northwest came from Michigan and the eastern colonies came from Pennsylvania. Figure from Chapman (1947).

Figure 2: Map describing the distribution and abundance of beavers in Ohio in 2012. Image taken from http://wildlife.ohiodnr.gov/species-and-habitats/species- guide-index/mammals/beaver

Figure 3: Beaver population estimates in Ohio from 1980-2011. Image taken from http://wildlife.ohiodnr.gov/species-and-habitats/species-guide-index/mammals/beaver

Dayton

Cincinnati

Figure 4: Locations of study sites in southwest Ohio.

25 m 40 m

120 m

24 m

25 m

120 m

120 m C 25 m 40 m 240 m

Figure 5: Schematic of transect designs. The blue lines represent the water’s edge while the black lines represent the primary (120 m, 240 m) and secondary (25 m) transects. The green represents the forest vegetation. Transect Type A was utilized when the forested area on both sides of the stream channel was >25 m. Transect Type B was utilized when only one forested bank was > 25 m and the other bank had a tree/shrub layer that was < 25 m. Transect Type C was utilized for pond/ lake sites with > 25 m forested vegetation. 27

Fig ure 6: Photos taken during summer 2018 from Browning (Model BTC-5HDE) trail cameras placed at Bachelor Pond (Oxford, OH).

Proportion Cut

Figure 7: Barplot of proportion of stems cut out of all stems within each distance class.

Distance refers to how far the stem was from the water’s edge.

Diameter Class (mm)

Figure 8: Barplot of proportion of stems cut out of all stems within each diameter class

Figure 9: Barplot of proportion of L. maackii stems cut out of all L. maackii stems within each distance class. Distance refers to how far the stem was from the water’s edge.

Diameter Class (mm)

Figure 10: Barplot of proportion of L. maackii stems cut out of all L. maackii stems within each diameter class.

Figure 11: Barplot of proportion of cut stems belonging to each genus.

Figure 12: Electivity values (Ei) for all genera found within this study. The blue points are native genera while the red points are invasive genera. The black x’s represent the mean electivity for each genus and the black line represents an electivity of 0. The size of the point corresponds with the number of sites that shared the same electivity value for that genus.

Figure 13: Electivity values (Ei) for genera found within five or more sites. The blue points are native genera while the red points are invasive genera. The black x’s represent the mean electivity for each genus and the black line represents an electivity of 0. The size of the point corresponds with the number of sites that shared the same electivity value for that genus.

2 Figure 14: Regressions of L. maackii electivity (Ei) on (A) average canopy openness (p = 0.158, R = 0.186), (B) beaver residency time (years) per site (p = 0.156, R2 = 0.189), and (C) density of preferred stems (p = 2 0.071, R = 0.351), among the eight sites. Site abbreviations can be found in Table 1. 36

Appendix 1

The total number of stems, stems cut, proportion cut, proportion of total stems cut and electivity (Ei) for each genus at each site.

BAWO Common Name(s) Genera Total Stems Prop. Prop. Of Ei Stems Cut Cut Cut Black Willow Salix 7 7 1.00 0.09 0.57 Border privet Ligustrum 9 8 0.89 0.11 0.53 Flowering/silky Cornus 7 6 0.86 0.08 0.52 dogwood Autumn olive Elaeagnus 21 12 0.57 0.16 0.35 Box Elder Acer 4 2 0.50 0.03 0.29 Burning bush Euonymus 2 1 0.50 0.01 0.29 Black Cherry Prunus 6 3 0.50 0.04 0.29 American/Slippery Ulmus 8 4 0.50 0.05 0.29 Elm Blue/green ash Fraxinus 17 6 0.35 0.08 0.13 Red/sugar maple Acer 13 4 0.31 0.05 0.06 American Sycamore Platanus 14 3 0.21 0.04 -0.12 Grapevine Vitis 5 1 0.20 0.01 -0.16 Chinkapin/Red Oak Quercus 6 1 0.17 0.01 -0.24 Black Locust Robinia 6 1 0.17 0.01 -0.24 Amur honeysuckle Lonicera 284 16 0.06 0.21 -0.66 Eastern redcedar Juniperus 19 1 0.05 0.01 -0.68 Honey locust Gleditsia 2 0 0.00 0.00 -1.00 Sweetgum Liquidambar 1 0 0.00 0.00 -1.00 Tulip Poplar Liriodendron 1 0 0.00 0.00 -1.00 Osage Orange Maclura 3 0 0.00 0.00 -1.00 Mulberry Morus 1 0 0.00 0.00 -1.00 Norway Spruce Picea 1 0 0.00 0.00 -1.00 Big tooth aspen Populus 2 0 0.00 0.00 -1.00 American Basswood Tilia 1 0 0.00 0.00 -1.00 Total 440 76 ACLA Box Elder Acer 1 1 1.00 0.02 0.73 Red/Sugar Maple Acer 28 7 0.25 0.15 0.23 American/Slippery Ulmus 12 3 0.25 0.06 0.23 Elm Autumn Olive Elaeagnus 21 5 0.24 0.10 0.21 Bitternut/Shagbark Carya 5 1 0.20 0.02 0.13 Hickory Blue/green Ash Fraxinus 10 2 0.20 0.04 0.13 Amur Honeysuckle Lonicera 103 20 0.19 0.42 0.11 37

Chinkapin/Red Oak Quercus 34 6 0.18 0.13 0.07 Black Cherry Prunus 13 2 0.15 0.04 0.00 Flowering Dogwood Cornus 8 1 0.13 0.02 -0.11 Ohio Buckeye Aesculus 10 0 0.00 0.00 -1.00 Musclewood Carpinus 2 0 0.00 0.00 -1.00 Euonymous Vine Euonymus 4 0 0.00 0.00 -1.00 American Beech Fagus 29 0 0.00 0.00 -1.00 Osage Orange Maclura 1 0 0.00 0.00 -1.00 Eastern Cottonwood Populus 2 0 0.00 0.00 -1.00 American Basswood Tilia 7 0 0.00 0.00 -1.00 Grapevine Vitis 2 0 0.00 0.00 -1.00 Total 292 48 HUNO Chinkapin/Red/White Quercus 4 2 0.50 0.13 0.72 Oak Blue/Green Ash Fraxinus 11 5 0.45 0.31 0.69 Grapevine Vitis 12 4 0.33 0.25 0.61 Box Elder Acer 21 3 0.14 0.19 0.27 American/Slippery Ulmus 16 2 0.13 0.13 0.21 Elm Sugar maple Acer 26 0 0.00 0.00 -1.00 Ohio Buckeye Aesculus 1 0 0.00 0.00 -1.00 Musclewood Carpinus 3 0 0.00 0.00 -1.00 Hackberry Celtis 7 0 0.00 0.00 -1.00 Eastern Redbud Cercis 6 0 0.00 0.00 -1.00 Flowering Dogwood Cornus 3 0 0.00 0.00 -1.00 Autumn Olive Elaeagnus 1 0 0.00 0.00 -1.00 American Beech Fagus 34 0 0.00 0.00 -1.00 Honey Locust Gleditsia 1 0 0.00 0.00 -1.00 Black Walnut Juglans 1 0 0.00 0.00 -1.00 Amur Honeysuckle Lonicera 36 0 0.00 0.00 -1.00 American Sycamore Platanus 10 0 0.00 0.00 -1.00 Tulip Poplar Populus 3 0 0.00 0.00 -1.00 Black Cherry Prunus 3 0 0.00 0.00 -1.00 Total 199 16 MIWI Autumn Olive Elaeagnus 30 30 1.00 0.23 0.53 Grapevine Vitis 2 2 1.00 0.02 0.53 Box Elder Acer 11 7 0.64 0.05 0.35 Hackberry Celtis 2 1 0.50 0.01 0.24 Amur Honeysuckle Lonicera 227 87 0.38 0.66 0.11 Blue/Green Ash Fraxinus 3 1 0.33 0.01 0.04 Ohio Buckeye Aesculus 16 2 0.13 0.02 -0.42 Black Walnut Juglans 22 2 0.09 0.02 -0.74 Flowering Dogwood Cornus 1 0 0.00 0.00 -1.00

Honey Locust Gleditsia 2 0 0.00 0.00 -1.00 Eastern Redcedar Juniperus 1 0 0.00 0.00 -1.00 American Sycamore Platanus 3 0 0.00 0.00 -1.00 Black Cherry Prunus 3 0 0.00 0.00 -1.00 Total 323 132 OXLA Silky Dogwood Cornus 1 1 1.00 0.01 0.47 Black Willow Salix 1 1 1.00 0.01 0.47 American/Slippery Ulmus 2 2 1.00 0.02 0.47 Elm Box Elder Acer 25 17 0.68 0.19 0.31 Green/Blue Ash Fraxinus 17 9 0.53 0.10 0.19 Osage Orange Maclura 2 1 0.50 0.01 0.16 Callery Pear Pyrus 2 1 0.50 0.01 0.16 Autumn Olive Elaeagnus 26 8 0.31 0.09 -0.08 Black Cherry Prunus 9 2 0.22 0.02 -0.24 American Sycamore Platanus 20 4 0.20 0.04 -0.29 Amur Honeysuckle Lonicera 234 43 0.18 0.48 -0.32 Hackberry Celtis 1 0 0.00 0.00 -1.00 Eastern Redbud Cercis 1 0 0.00 0.00 -1.00 Burning Bush/Vine Euonymus 2 0 0.00 0.00 -1.00 Sweetgum Liquidambar 2 0 0.00 0.00 -1.00 Black Locust Robinia 1 0 0.00 0.00 -1.00 Viburnum Viburnum 1 0 0.00 0.00 -1.00 Total 347 89 PEPK Box Elder Acer 1 1 1.00 0.02 0.74 Grapevine Vitis 2 1 0.50 0.02 0.54 Autumn Olive Elaeagnus 19 8 0.42 0.18 0.48 Green Ash Fraxinus 3 1 0.33 0.02 0.38 Amur Honeysuckle Lonicera 232 33 0.14 0.75 -0.03 Hackberry Celtis 3 0 0.00 0.00 -1.00 Eastern Redbud Cercis 1 0 0.00 0.00 -1.00 Hawthorn Crataegus 2 0 0.00 0.00 -1.00 Honey Locust Gleditsia 3 0 0.00 0.00 -1.00 Black Wlanut Juglans 2 0 0.00 0.00 -1.00 Eastern Redcedar Juniperus 15 0 0.00 0.00 -1.00 Osage Orange Maclura 6 0 0.00 0.00 -1.00 American Sycamore Platanus 4 0 0.00 0.00 -1.00 Black Cherry Prunus 2 0 0.00 0.00 -1.00 Black Locust Robinia 5 0 0.00 0.00 -1.00 Slippery Elm Ulmus 1 0 0.00 0.00 -1.00 Total 301 44 39

POCR Black Willow Salix 3 1 0.33 0.01 0.51 American Elm Ulmus 1 1 1.00 0.01 0.51 Box Elder Acer 24 23 0.96 0.27 0.49 Autumn Olive Elaeagnus 54 43 0.80 0.51 0.42 Black Locust Robinia 3 1 0.33 0.01 0.01 Eastern Redcedar Juniperus 8 2 0.25 0.02 -0.13 American Sycamore Platanus 16 2 0.13 0.02 -0.44 Hawthorn Crataegus 137 6 0.04 0.07 -0.76 Amur Honeysuckle Lonicera 112 5 0.04 0.06 -0.76 American Beech Fagus 1 0 0.00 0.00 -1.00 Honey Locust Gleditsia 1 0 0.00 0.00 -1.00 Osage Orange Maclura 3 0 0.00 0.00 -1.00 Red Oak Quercus 3 0 0.00 0.00 -1.00 Grapevine Vitis 1 0 0.00 0.00 -1.00 Total 367 84 WEGA Silky Dogwood Cornus 4 4 1.00 0.22 0.52 Box Elder Acer 5 3 0.60 0.17 0.31 Amur Honeysuckle Lonicera 18 5 0.28 0.28 -0.07 Green Ash Fraxinus 5 1 0.20 0.06 -0.23 Silver Maple Acer 32 5 0.16 0.28 -0.34 Eastern Cottonwood Populus 1 0 0.00 0.00 -1.00 American Elm Ulmus 4 0 0.00 0.00 -1.00 Total 69 18

Appendix 2 – R code and output for binomial generalized regressions and linear regressions Binomial Generalized Regression – All genera > playmix <- glmer(data=Play_Data_Beaver_Browse, Cut ~ Genus + DistancetoWater * Diameter+ (1|Site), family="binomial") > Anova(playmix, type = 3) Analysis of Deviance Table (Type III Wald chisquare tests) Response: Cut Chisq Df Pr(>Chisq) (Intercept) 20.0208 1 7.66e-06 *** Genus 291.3942 24 < 2.2e-16 *** DistancetoWater 11.7042 1 0.0006236 *** Diameter 46.7633 1 8.01e-12 *** DistancetoWater:Diameter 0.0007 1 0.9789110 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > fixef(playmix) (Intercept) GenusAesculus 1.893129e+00 -3.896227e+00 GenusCeltis GenusCercis -1.355395e+00 -1.596915e+01 GenusCornus GenusCrataegus -9.022747e-02 -4.483286e+00 GenusElaeagnus GenusEuonymus -3.595955e-01 -2.529353e+00 GenusFagus GenusFraxinus -1.748436e+01 -3.781055e-01 GenusGleditsia GenusJuglans -1.491821e+01 -2.196334e+00

GenusJuniperus GenusLonicera -1.667927e+00 -2.529749e+00 GenusMaclura GenusPlatanus -1.002139e+00 -2.214005e+00 GenusPopulus GenusPrunus -1.257890e+01 -1.753344e+00 GenusQuercus GenusRobinia -5.893130e-01 -1.262313e+00 GenusSalix GenusUlmus 1.756958e+01 7.881699e-02 GenusUncommon GenusVitis -7.935103e-01 -5.086370e-01 DistancetoWater Diameter -6.682923e-02 -1.989350e-02 DistancetoWater:Diameter 9.907931e-06 > ranef(playmix) $Site (Intercept) ACLA -0.52774548 BAWO -0.41238982 HUNO -1.66674086 MIWI 1.30876070 OXLA -0.19477887 PEPK 0.08274149 POCR 0.52080197 WEGA 0.94319472

Binomial Generalized Regression with Genera from 5+ sites > playmix2 <- glmer(data=Play_Data_Beaver_Browse_4_Uncommon, Cut ~ Genus + DistancetoWater * Diameter+ (1|Site), family="binomial") > Anova(playmix2, type = 3) Analysis of Deviance Table (Type III Wald chisquare tests) Response: Cut Chisq Df Pr(>Chisq) (Intercept) 23.3582 1 1.345e-06 *** Genus 276.4757 13 < 2.2e-16 *** DistancetoWater 11.2990 1 0.0007755 *** Diameter 41.9363 1 9.430e-11 *** DistancetoWater:Diameter 0.0014 1 0.9696411 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > fixef(playmix2) (Intercept) GenusCercis 2.215434e+00 -1.336159e+01 GenusCornus GenusCrataegus -5.192742e-01 -4.766465e+00 GenusElaeagnus GenusFraxinus -6.713375e-01 -7.440534e-01 GenusGleditsia GenusLonicera -1.515272e+01 -2.821226e+00 GenusMaclura GenusPlatanus -1.403234e+00 -2.515036e+00 GenusPrunus GenusUlmus -2.063601e+00 -3.720282e-01 GenusUncommon GenusVitis

-1.777434e+00 -8.388656e-01 DistancetoWater Diameter -6.700504e-02 -1.897518e-02 DistancetoWater:Diameter -1.484738e-05 > ranef(playmix2) $Site (Intercept) ACLA -0.51772001 BAWO -0.28013066 HUNO -1.73823082 MIWI 1.11687152 OXLA -0.25015861 PEPK 0.02691531 POCR 0.45208016 WEGA 1.24936635

Binomial Generalized Regression- just Lonicera > playmix3 <- glmer(data=Play_Data_Beaver_Browse_HS_Only, Cut ~ DistancetoWater * Diameter+ (1|Site), family="binomial") > Anova(playmix3, type = 3) Analysis of Deviance Table (Type III Wald chisquare tests) Response: Cut Chisq Df Pr(>Chisq) (Intercept) 0.8711 1 0.35065 DistancetoWater 24.3986 1 7.833e-07 *** Diameter 29.6135 1 5.274e-08 *** DistancetoWater:Diameter 10.8096 1 0.00101 **

--- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 > fixef(playmix3) (Intercept) DistancetoWater 0.543339700 -0.168109734 Diameter DistancetoWater:Diameter -0.053481986 0.002343947 > ranef(playmix3) $Site (Intercept) ACLA 0.3286117 BAWO -0.9052624 HUNO -1.9521513 MIWI 1.4162520 OXLA 0.2007755 PEPK 0.5333624 POCR -0.8482040 WEGA 1.5657871

Linear Regressions: L. maackii Ei vs Canopy Openness Call: lm(formula = Ei ~ AveCanOp, data = HSEiCanopy) Residuals: Min 1Q Median 3Q Max -0.48691 -0.22509 0.00099 0.12633 0.62022 Coefficients: Estimate Std. Error t value Pr(>|t|)

(Intercept) -0.51799 0.18074 -2.866 0.0286 * AveCanOp 0.01690 0.01048 1.613 0.1580 --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.3869 on 6 degrees of freedom Multiple R-squared: 0.3024, Adjusted R-squared: 0.1861 F-statistic: 2.6 on 1 and 6 DF, p-value: 0.158

L. maackii Ei vs # Preferred Stems per m2 Call: lm(formula = PreferredvsHSEI$LoniceraEi ~ PreferredvsHSEI$PropCut) Residuals: Min 1Q Median 3Q Max -0.62562 -0.13414 0.02657 0.18657 0.45145 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 6.394e-04 1.933e-01 0.003 0.9975 PreferredvsHSEI$PropCut -2.206e+01 1.008e+01 -2.189 0.0712 (Intercept) PreferredvsHSEI$PropCut . --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.3454 on 6 degrees of freedom Multiple R-squared: 0.444, Adjusted R-squared: 0.3514 F-statistic: 4.792 on 1 and 6 DF, p-value: 0.07116

L. maackii Ei vs Beaver Residency Time (Years)

Call: lm(formula = ResTimeEi$Ei ~ ResTimeEi$ResTime) Residuals: Min 1Q Median 3Q Max -0.44548 -0.21812 -0.07346 0.16531 0.64732 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -0.05312 0.21741 -0.244 0.815 ResTimeEi$ResTime -0.03228 0.01990 -1.622 0.156 Residual standard error: 0.3862 on 6 degrees of freedom Multiple R-squared: 0.3048, Adjusted R-squared: 0.1889 F-statistic: 2.631 on 1 and 6 DF, p-value: 0.1559

Appendix 3 Barplot of the proportion of L. maackii stems cut within each size class for stems close to the water’s edge and stems far from the water’s edge. Close and far stems were determined by the median in the dataset. All stems below or equal to the median of 3.15 m were considered close while all stems above 3.15 m were considered far.