Northeastern Naturalist 174 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 22001166 NORTHEASTERN NATURALIST 2V3(o1l). :2137,4 N–1o8. 31 A Case for Unique Habitat Selection by Sigara mathesoni (Hemiptera: Corixidae) in South-Central Pennsylvania Dustin R. Shull1, Richard L. Stewart Jr.2, Todd M. Hurd2, and Theo Light2,* Abstract - Sigara mathesoni (Hemiptera: Corixidae), which predominantly inhabits the northern United States and Canada, has recently been found in carbonate streams with stable and cold temperature regimes in south-central Pennsylvania. The focus of this research was to determine if there is a difference in occurrence of S. mathesoni between carbonate and non-carbonate streams, and whether abiotic factors within these habitats can be used to explain or predict its regional distribution. Our results suggest that S. mathesoni predominantly occurs in carbonate spring habitats, and that parameters such as alkalinity and daytime summer water temperature may best predict occurrence. This work validates observations in other literature and supports the hypothesis that S. mathesoni is a rare species in Pennsylvania and possibly throughout its southern range. Introduction The seminal papers concerning Corixidae (water boatmen) were published between 1930 and 1950, yet only sporadic research on various species has been conducted since. Sigara mathesoni (Hungerford) (Hemiptera: Corixidae) mainly inhabits regions of southern Canada and the northern United States. However, evidence suggests clustered distribution reaching as far south as Missouri, Indiana, Ohio, and Pennsylvania (Chordas et al. 2002, Froeschner 1962, Harp et al. 2008, Shull et al. 2010). Complete information about life history and distribution within Pennsylvania are still lacking for numerous corixid species, including S. mathesoni. In south-central Pennsylvania, S. mathesoni may only occur and reproduce in streams with cold temperature regimes, as none were discovered in warmer-water streams located close to the sites previously reported (Shull et al. 2010). Generally, corixids reside in shallow, slow-moving areas of streams (Tolonen et al. 2003) and, like many other aquatic species, may require diel temperature fluctuations of greater than 10 °C to begin reproduction (Barahona et al. 2005, Sweeney et al. 1977). Therefore, the upper reaches of cold-water carbonate streams, where temperatures fluctuate only a few degrees annually, likely exclude viable populations of more common corixid species. As a result, S. mathesoni may occupy a specialized niche within its southern range. The focus of this research is to more fully document the distribution of S. mathesoni in carbonate and non-carbonate streams and determine whether abiotic factors can be used to predict its regional distribution. 1Pennsylvania Department of Environmental Protection, Harrisburg, PA 17105. 2Shippensburg University of Pennsylvania, Shippensburg, PA 17257. *Corresponding author - tsligh@ship.edu. Manuscript Editor: Daniel Pavuk Northeastern Naturalist Vol. 23, No. 1 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 175 Field Site Description The Great Valley and nearby areas within south-central Pennsylvania (Franklin and Cumberland counties) were chosen based on the density of carbonate streams. These cold-water carbonate streams are rare compared to the more common noncarbonate streams of Pennsylvania (approximately 800 vs. 86,000 stream miles, respectively; Botts 2009). Carbonate-stream distribution is clustered in certain regions of PA with geology rich in limestone and dolomite. The Ordovician and Cambrian formations of the Great Valley are representative of such geological characteristics (Becher and Root 1981). In Pennsylvania, Ordovician and Cambrian formations stretch from lower Franklin County northeast to Northampton County. There are also smaller deposits concentrated in Centre, Mifflin, Adams, York, and Lancaster counties, PA (DCNR 2007). Hungerford (1948) reported the occurrence of S. mathesoni in the vicinity of 3 Pennsylvanian towns: State College (21 November 1938), Bethlehem (14 May 1912), and Wyomissing (27 August1934). All of these observations, as well as those of Shull et al. (2010) at Big Spring Creek, fall within the Ordovician and Cambrian carbonate formations. This distribution may suggest that abiotic stream characteristics such as alkalinity, conductivity, and water temperature can predict the regional distribution of S. mathesoni. Prior research based on stream observations and geologic distribution also supports this hypothesis (Chordas et al. 2002). However, Chordas et al. (2002) focused on documenting new species in Indiana, Michigan, Ohio, and Pennsylvania, and did not record abiotic conditions. It’s important to note that not all streams that fall within carbonate geology are classified as carbonate streams (Botts 2009). Therefore, certain parameters must be established to distinguish between the 2 stream types. Methods For this study, 2 different stream types (carbonate and non-carbonate) were chosen to compare the differences in S. mathesoni occurrence. We selected carbonate stream sites based on the carbonate criteria described by Botts (2009), where alkalinity must be at or above 140 mg/L CaCO3, stream temperature must remain around 4–18 °C year round, and stream size must be small (drainage area less than 50 km2) with little to no surface-water inputs. Specifically, we sampled 9 carbonate streams and 11 non-carbonate streams (Table 1) throughout south-central Pennsylvania. For simplicity, we classified sites that were determined to be carbonate influenced (having most, but not all carbonate characteristics) as non-carbonate. We collected specimens using a truncated D-framed net with 500-micron mesh. The sampling process at each site consisted of 20 jabs (~1 m long) into aquatic vegetation and sediment, focusing on the best available habitat for corixids (Tolonen et al. 2003). We conducted species identification of S. mathesoni in the field using a 10x field lens to observe the distinct presence of a longitudinal brown stripe in the intraocular space (Hungerford 1948), and took vouchers from each site for verification. We then determined relative abundance of nymphal and adult S. mathesoni for Northeastern Naturalist 176 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 Vol. 23, No. 1 Table 1. Sampling results of 9 carbonate and 11 non-carbonate streams in south-central Pennsylvania that were sampled for Sigara mathesoni (S. m.). Specific Site Stream Ambient DO conductance Alkalinity S. m. Other code Stream name Stream type temp (°C) temp (°C) (mg/L) pH (μS/cm @ 25°C) (mg/L CaCO3) total Corixidae C1 Big Spring Creek Carbonate 12.1 29.0 10.50 7.49 491 135 28 No C2 Alexander Spring Creek Carbonate 12.0 26.7 10.05 7.42 640 186 1 Yes C3 Hogestown Run Carbonate 18.1 30.0 12.23 7.93 627 164 2 Yes C4 Letort Spring Run Carbonate 12.4 29.4 10.67 7.63 570 160 0 Yes C5 Trout Run Carbonate 14.7 28.9 10.04 7.85 583 148 2 No C6 Green Spring Creek Carbonate 15.4 32.2 9.65 7.53 688 191 7 Yes C7 Mount Rock Spring Carbonate 12.2 32.2 10.70 7.46 612 171 52 Yes C8 Middle Spring Creek Carbonate 15.5 31.0 12.02 7.96 368 102 391 Yes C9 Falling Spring Creek Carbonate 13.3 29.4 11.80 7.64 620 179 3 No N1 Birch Run Non-carbonate 17.8 29.0 9.36 4.98 17 1 0 No N2 Conewago Creek Non-carbonate 20.7 24.0 9.09 7.33 76 27 0 Yes N3 Conodoguinet Creek Non-carbonate 22.1 29.0 9.33 7.26 82 21 0 No N4 Mountain Creek Non-carbonate 24.4 29.0 8.50 6.44 47 16 0 No N5 Muddy Run Non-carbonate 16.1 22.0 8.63 7.85 638 218 12 Yes N6 Opossum Creek Non-carbonate 20.3 24.0 9.17 7.37 129 29 0 No N7 Plum Run Non-carbonate 22.9 26.0 8.83 7.62 503 161 0 Yes N8 Reservoir Hollow Run Non-carbonate 21.4 29.0 8.52 6.40 78 27 0 No N9 Rocky Mountain Creek Non-carbonate 23.7 30.0 8.55 7.34 106 25 0 No N10 South Branch Conodoguinet Non-carbonate 21.3 29.0 10.66 6.90 57 18 0 No N11 West Branch Conococheague Non-carbonate 20.8 28.0 8.60 7.98 459 164 0 No Northeastern Naturalist Vol. 23, No. 1 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 177 each site. Sampling occurred during July and August when late instars and adults of S. mathesoni were most likely to co-occur. The presence of adults and late instars indicated an established population as well as potential reproduction. We also recorded the occurrence of other corixid species at each site, but did not identify to species or count for relative abundance those specimens. We field-measured 5 physicochemical parameters at each site: water temperature, conductivity, and pH with an Oakton® CON 6 digital meter; dissolved oxygen with a YSI® model 550A digital meter; and alkalinity with the Hach® model 16900 digital titrator. Although we sampled only once at each site, we took all water temperature readings near mid-day (1100–1500 hours) in July or August, so they were representative of some of the highest annual temperatures in each stream. Data were analyzed using R version 3.1.2 software (R Core Team 2015). We conducted a Fisher’s exact test to determine if there was an overall difference between S. mathesoni and other corixid occurrences in carbonate versus noncarbonate streams. To determine which physiochemical factors were most closely associated with the presence of S. mathesoni, we employed logistic regression and information-theoretic model selection using AICc (Burnham and Anderson 2002). We developed a set of candidate models that included all possible linear combinations of 1–3 of the 5 water-quality parameters, excluding combinations of highly correlated (r > 0.75) measures. The only excluded combination was of alkalinity and specific conductance (r = 0.98). A global model containing all 5 parameters was also included. Using the R package AICcmodavg (Mazerolle 2013), we ranked all candidate models based on AICc values, and based inference on a confidence set of the top-ranked models. Results We found a total of 486 S. mathesoni in 8 of 9 carbonate streams and 12 in 1 of 11 non-carbonate streams (Table 1, Fig. 1). Sigara mathesoni occurrence was strongly associated with carbonate-classified streams (Fisher’s exact test: P < 0.001). Other corixids were also more frequently observed in carbonate streams (Table 1), but differences in occurrence between the 2 stream types were not significant (Fisher’s exact test: P = 0.17). Model selection resulted in a large number of similarly ranked models that had similar predictive ability (Table 2). Seven models had delta AICc values of ~2 or below, and of these, all the 2-parameter models had predictive success of 90 or 95% (most misclassified only the Letort Spring Run, predicting S. mathesoni would be present in the one carbonate stream in which it was not found). The most successful models typically included a combination of specific conductance, alkalinity or pH, along with either daytime summer water temperature or dissolved oxygen (Table 2). The models predicted presence of S. mathesoni in sites with alkalinity above 100 mg/L as CaCO3, and daytime summer water temperatures 18 °C or below, consistent with the definition of a carbonate stream (Fig. 2). Other corixids did not show a consistent relationship with these 2 parameters (Fig. 2). Northeastern Naturalist 178 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 Vol. 23, No. 1 Discussion This study clarifies several characteristics of S. mathesoni distribution in southcentral Pennsylvania. Sigara mathesoni does occur with much greater frequency in carbonate streams than non-carbonate streams, and their presence in this small Figure 1. Sampling locations and geographic distribution of positive and negative S. mathesoni occurrence in carbonate and non-carbonate streams. Sites are labeled with individual site code shown in Table 1. Table 2. Ninety-five percent confidence set of models predicting presence/absence of S. mathesoni. All models are logistic regression models built using the glm() function in R, with family set to binomial and the link function to “logit”. Models were ranked using the AICcmodavg package (Mazerolle 2013). Model parameters K AICc ΔAICc AICc weight Predictive success cond+DO 3 15.999 0.000 0.164 0.95 alk+DO 3 16.322 0.323 0.140 0.95 cond 2 16.962 0.963 0.101 0.85 cond+temp 3 17.272 1.273 0.087 0.90 pH+temp 3 17.525 1.527 0.077 0.95 temp 2 17.809 1.810 0.066 0.90 alk+temp 3 18.035 2.037 0.059 0.95 cond+pH+DO 4 18.950 2.951 0.038 0.95 cond+temp+DO 4 19.068 3.070 0.035 0.95 alk+temp+DO 4 19.187 3.188 0.033 0.95 cond+pH+temp 4 19.431 3.432 0.030 0.95 alk+pH+DO 4 19.463 3.464 0.029 0.95 temp+DO 3 19.596 3.598 0.027 0.90 cond+pH 3 19.618 3.619 0.027 0.80 alk 2 19.843 3.844 0.024 0.80 alk+pH+temp 4 20.179 4.180 0.020 0.95 Northeastern Naturalist Vol. 23, No. 1 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 179 sample can be predicted accurately using combinations of any of a number of commonly measured water-quality parameters. Specific conductance and dissolved oxygen were the best indicators for S. mathesoni occurrence in this dataset, but may not be as reliable across a more diverse set of streams. For example, specific conductance can be high in streams for reasons other than the presence of a carbonate influence, such as a forested headwater stream with acid mine drainage. Dissolved oxygen can vary a great deal within sites, and can undergo major diel fluctuations in productive sites, even those with quite stable temperatures. Alkalinity and daytime summer water temperature make the most sense as parameters to predict the Figure 2. Presence and absence of S. mathesoni and other corixids in relationship to alkalinity and daytime summer water temperature. Northeastern Naturalist 180 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 Vol. 23, No. 1 occurrence of S. mathesoni, as these are the 2 most important parameters for distinguishing between carbonate and non-carbonate streams (Botts 2009). For sites where alkalinity has not been assessed, a combination of specific conductance, pH, and daytime summer water temperature is likely to produce an acceptable prediction of S. mathesoni presence as well (Table 2). The association of S. mathesoni with carbonate streams may be largely due to their stable, cool temperature regimes. This study did not compare diel or annual temperature variability due to a lack of resources, but this factor would be interesting to assess in future research. Because we sampled relatively few coldtemperature streams that lacked a high carbonate influence, it is difficult to assess the independent importance of alkalinity or specific conductance as predictors of S. mathesoni presence, though one or the other of these parameters was present in most of the top models (Table 2). Given these findings, underlying geology may assist in rough reconnaissance of stream sites likely to support S. mathesoni, particularly when associated with larger carbonate streams (Fig. 1). Many carbonate streams had high densities of aquatic vegetation and heavy sedimentation. Sedimentation in carbonate streams is somewhat natural, but can be exacerbated due to anthropogenic effects (nutrient loading, quarrying, and riparian disturbance). It is these characteristics, however, that provide optimal habitat for corixids (Tolonen et al. 2003). Non-carbonate streams, on the other hand, often had higher velocities associated with high gradients, and much less in-stream vegetation. When compared to carbonate streams, it is clear that the non-carbonate streams we sampled did not provide abundant corixid habitat. The lack of suitable habitat alone may explain why more corixids were discovered in carbonate springs, but it does not adequately explain the specific species/habitat selection of S. mathesoni observed during the study. Several site-specific results were noteworthy and are expanded upon here. Letort Spring Run was the only carbonate stream where S. mathesoni was not captured. We believe this may be a false-negative finding for a variety reasons. First, Letort Spring Run was one of the largest true carbonate streams visited, with several spring origins, making successful dip netting much more difficult per unit area. Therefore, it may be appropriate to weight dip-net attempts according to stream size. Additionally, the high abundance of aquatic and riparian vegetation and lack of a distinct bank made it difficult to sample appropriate habitat. In this instance and for future work, light traps may be necessary to accurately assess streams with these characteristics (Hungerford et al. 1955). Middle Spring Creek had, by far, the greatest abundance of S. mathesoni (391) out of all sampled streams. The reason for this is unclear, but fish ponds that were connected to the stream origin just upstream of the sampling site may have played a role. These ponds are a significant nutrient source in the form of fish/duck food and waste. Like many other opportunistic invertebrates, corixids may feed off the abundant bacterial/algal bio-films and fine particulate organic matter associated with hatchery waste (Hurd et al. 2008). This idea is further supported by Shull (2011) visually confirming S. mathesoni consuming filamentous algae during laboratory experiments. Northeastern Naturalist Vol. 23, No. 1 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 181 Muddy Run was the only non-carbonate stream where S. mathesoni were captured. It is not considered a true carbonate stream because of variable discharge during rain events, surface-water influences, and lack of flora and fauna associated with carbonate streams (Botts 2009). Nonetheless, the high pH, specific conductance, and alkalinity in this stream, along with its relatively cool temperature (Table 1), suggest that it is highly influenced by carbonate springs. Indeed, Muddy Run was sampled below the confluence with Rowe Run, a carbonate spring. Therefore, one explanation for S. mathesoni occurrence in Muddy Run is that they were simply swept in from Rowe Run. It is also possible that a few transient S. mathesoni flew in from another nearby carbonate stream. Muddy Run had little vegetation to slow water velocity; however, it is interesting to note that sand bars created the necessary pools where fine particulate organic material/detritus collected. It was in these sand-bar pools where the majority of S. mathesoni were captured. While these observations help explain the distribution of S. mathesoni, there are numerous questions left unanswered. For instance, interspecific competition has been shown to play an important role in the distribution of rock-pool corixids (Pajunen and Pajunen 1993). This could also be a factor influencing S. mathesoni distribution, particularly if the stable, cool thermal conditions of these carbonate springs are limiting to other corixid species. Given the distribution-analysis results, carbonate streams seem to provide a unique habitat for S. mathesoni. Some characteristics of these streams, such as temperature regime, do not meet the reproductive requirements of other species within the same genus (Sweeney and Schnack 1977). It is possible S. mathesoni has adapted to be a cold-spring specialist by not requiring diel water temperature fluctuations during critical times of year to reproduce. This trait would allow it to colonize areas unavailable to other corixids. Another possibility is that S. mathesoni is adapted to colder temperatures, whereas other species simply need higher temperatures for egg development or instar growth. The distribution of S. mathesoni within carbonate streams of Pennsylvania may be understood by an analysis of the greater distribution of the species. Sigara mathesoni is largely a northerly species that is obviously adapted to colder regimes. As its range projects farther south, cold regimes become rare with the exception of areas that contain karst geology and cold-water habitats (Froeschner 1962). Given that there are only about 800 stream miles with these criteria within Pennsylvania, it is possible that S. mathesoni is a rare species at least within the confines of the Commonwealth and may deserve special protection. More research into S. mathesoni distribution should be completed to refine these findings and to ascertain the true range of this corixid. Acknowledgments We sincerely thank Dr. Gregory Paulson for graciously providing feedback throughout the course of this study, providing internal review of this manuscript, and for allowing the use of his laboratory and equipment. We are also grateful to Dr. Steve Chordas III for confirming identification of Sigara mathesoni nearly 6 years ago, which officially began this Northeastern Naturalist 182 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 Vol. 23, No. 1 venture. Thanks also goes to William Botts for providing the list of true carbonate streams in the area, which saved an immeasurable amount of time in the field. Field data collection would not have been possible without the tireless efforts and knowledge of Barry Myers and Scott Jones. Literature Cited Barahona, J., A. Millan, and J. Velasco. 2005. Population dynamics, growth, and production of Sigara selecta (Fieber, 1848) (Hemiptera, Corixidae) in a Mediterranean hypersaline stream. Freshwater Biology 50:2101–2113. Becher, A., and S. Root. 1981, Ground water and Geology of the Cumberland Valley, Cumberland County, Pennsylvania. Pennsylvania Geological Survey. Available online at www.dcnr.state.pa.us/topogeo/pub/water/w050.aspx. Accessed 15 July 2010. Botts, W. 2009. An index of biological integrity for “true” limestone streams. PA Department of Environmental Protection. Bureau of Point and Non-Point Source Management. Available online at http://www.portal.state.pa.us/portal/server.pt/community/water_ quality_standards/10556/technical_documentation_macroinvertebrate_stream_protocols/ 554005. Accessed 15 July 2010. Burnham, K.P., and D.R. Anderson. 2002. Model Selection and Multimodel Inference: A Practical Information-theoretic Approach. Springer, New York, NY. Chordas, S.W., III, E.G. Chapman, P.L. Hudson, M.A. Chriscinske, and R.L. Stewart Jr. 2002. New Midwestern state records of aquatic Hemiptera (Corixidae: Notonectidae). Entomological News 113:310–314. Froeschner, R.C. 1962. Contributions to a synopsis of the Hemiptera of Missouri, Part V. Hydrometridae, Gerridae,Veliidae, Saldidae, Ochteridae, Gelastocoridae, Naucoridae, Belostomatidae Nepidae, Notonectidae, Pleidae, Corixidae. American Midland Naturalist. 67(1):208–240. Harp, G., P. Harp, and S. McCord. 2008. Aquatic macroinvertebrates collected from thirtytwo Missouri Ozark streams. Journal of the Arkansas Academy of Science 62:61–74. Hungerford, H.B. 1948. The Corixidae of the Western Hemisphere (Hemiptera). University of Kansas Science Bulletin 32:1-827. Hungerford, H.B., P.J. Spangler, and N.A. Walker. 1955. Subaquatic light traps for insects and other animal organisms. Transactions of the Kansas Academy of Science. 58(3):387–407 Hurd, T.M., S. Jesic, J.L. Jerin, N.W. Fuller, and D. Miller Jr. 2008. Stable-isotope tracing of trout hatchery carbon to sediments and foodwebs of limestone spring creeks. Science of the Total Environment 405:161–172. Mazerolle, M.J. 2013. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c). R package version 1.35. Available online at http://CRAN.R-project.org/ package=AICcmodavg. Accessed 1 June 2014. Pajunen, V.I., and I. Pajunen. 1993. Competitive interactions limiting the number of species in rock pools: Experiments with Sigara nigrolineata. Oecologia 95(2):220–225. Pennsylvania Department of Conservation and Natural Resources (DCNR). 2007. Geologic Map of Pennsylvania. Available online at http://www.dcnr.state.pa.us/topogeo/maps/ map7.pdf. Accessed 15 July 2010. R Core Team. 2015. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at http://www.R-project.org/. Shull, D.R. 2011. Distribution and Ecology of Sigara mathesoni (Hungerford 1948) (Heteroptera: Corixidae) in south-central Pennsylvania. M.Sc. Thesis. Shippensburg University, Shippensburg, PA. 102 pp. Northeastern Naturalist Vol. 23, No. 1 D.R. Shull, R.L. Stewart Jr., T.M. Hurd, and T. Light 2016 183 Shull, D.R., R.L. Stewart Jr., and G.S. Paulson. 2010. Application of Dyar’s Law to life stages of Sigara mathesoni (Hungerford) (Heteroptera: Corixidae). Entomological News 121(5):469–474. Sweeney, B.W., and J.A. Schnack. 1977. Egg development, growth, and metabolism of Sigara alternata (Say) (Hemiptera: Corixidae) in fluctuating thermal environments. Ecology 58:265–277. Tolonen, K.T., H. Hämäläinen, I.J. Holopainen, K. Mikkonen., and J. Karjalainen. 2003. Body size and substrate association of littoral insects in relation to vegetation structure. Hydrobiologia 499:179–190.