Introduction

The distributions of animal species are determined by many things, including historical (biogeography), ecological (competitors, predators and parasites, food sources), and anthropogenic (development, overuse, invasives) factors. To make informed recommendations on how best to prevent the decline of threatened and endangered species, we must examine the influence of these factors on the species' distributions. This is especially true because the distributions of threatened and endangered species are very often limited in geographic or ecological extent.

The crucial first step is determining the entire range of habitats in which a rare species occurs and those in which it does not. Thus, when feasible, surveys for populations should be done in all potentially appropriate habitats so that informed conservation recommendations can be made. This is particularly true for many threatened or endangered insect species, because they are generally poorly studied, often overlooked in habitat assessments, and seasonal, so that surveys at inappropriate times will not detect their presence (Evers 1994, Wisconsin Dept. of Natural Resources 1999).

The Lake Huron locust (Trimerotropis huroniana Wlk.), threatened in both Michigan and Wisconsin, is an excellent example of a species where gathering detailed information about the distribution of populations has direct bearing on conservation recommendations. Our previous knowledge of the distribution of this species was based primarily on work by Hubbell (1929) and Otte (1984), who compiled the existing museum records and examined several populations in the field. Both studies indicated that the species is restricted to northern Great Lakes dune habitats. This was the basis for its listing as threatened in both states (Evers 1994, Wisconsin Dept. of Natural Resources 1999). The extent of these earlier surveys was very limited, for Hubbell, probably because of travel difficulties and for Otte, because the main focus of his work was on communication behavior (Otte 1970). Together they documented 20 populations (16 by Hubbell, at least 8 by Otte, with some over-lap) of the species in Ontario and Michigan. Ballard (1987; 1989; 1991a,b) did additional work in the late 1980s and early 1990s. He located several additional populations, including the first Wisconsin population (Ballard 1989). Occasional sampling in Michigan by Nature Conservancy personnel detected a few more populations. In total, before the start of this study, 25 different populations had been documented.

Otte (1984) noted that the southern limits of the locust appeared to have changed between Hubbell's survey and his. Along the Lake Huron shore in Michigan, the locust was no longer present at southern localities noted by Hubbell, but had been replaced by its congener, T. maritima (Harris). In addition, on the eastern Lake Michigan shoreline the locust appeared to have extended its range to the south. Ballard's studies did not address these apparent range changes. None of the studies specifically mention dune sites where the locust was not found. Comparing known absences with known occurrences can be important in determining the most important factors influencing the distribution of the locust.

The purpose of the current study was to fully document the range of the Lake Huron locust in Michigan by surveying as many dune habitats for populations as possible. The information gathered from these surveys was used to assess the influence of historical and recent influences on the distribution and conservation status of the species.

Methods

We surveyed for populations of the locust over the course of 3 summers (1996 to 98) following preliminary work in 1995 (and one incidental observation in 2002). We visited every dune habitat to which we could gain access within the state of Michigan. These locations were along the shores of Lake Michigan, Lake Huron and eastern Lake Superior to Pictured Rocks National Lakeshore. During 1995, scattered localities were surveyed in connection with an ongoing project on nesting success in the piping plover (Wemmer unpub. data). The remainder of the shoreline was covered in sections, always during the adult flight season so that locusts could be detected both visually and auditorily. The northern section of the Lake Huron shoreline along the Lower Peninsula was surveyed during 1996. The shores of Lake Michigan, Lake Huron and eastern Lake Superior in the Upper Peninsula were surveyed during 1997. The southern section of the Lake Huron shoreline along the Lower Peninsula was surveyed during 1998. The southern section of the shore of Lake Michigan, along the Lower Peninsula, was surveyed during 1998.

At each site we checked for the presence of the locust and estimated the relative size of the population by estimating how many individuals were seen. At each site, except those inside Pictured Rocks and Sleeping Bear Dunes National Lakeshores, vouchers were taken. These samples generally consisted of 2 males; females were collected only occasionally. These vouchers are deposited in the University of Michigan Museum of Zoology. Sample days were chosen to be mostly sunny, warm days, suitable for locust activity. At each location we surveyed most of the available habitat (with the exception of localities with many miles of dunes). Thus it is unlikely that populations were overlooked at localities where we recorded no locusts.

We surveyed many localities each sampling day, and thus limited our assessment of habitat variables to qualitative measures. At each locality we noted the dominant species of plants: particularly known hosts, dune specialists, and known invasive species. The extent of invasion was recorded as none, small amount or large amount. The extent of other disturbances was noted (e.g., foot traffic, off-road vehicle use, recreational use, known or suspected revegetation efforts). We noted other grasshopper species present, particularly banded-winged locusts (subfamily Oedipodinae). In general, we did not observe locust behaviors, instead focusing our limited on-site time to determining presence/absence and habitat characteristics.

Results

The general distribution of the Lake Huron Locust is very similar to that previously indicated by Hubbell (1929) and Otte (1984). Overall we located 64 populations of T. huroniana (Table 1)Table 1.Table 1., although defining the limits of some of the populations is difficult [we were unable to survey several populations previously recorded by Otte (1984) and Hubbell (1929), especially those on islands]. The locust occurs on most suitable beaches in the northern Great Lakes, including the northern shores of Lakes Michigan and Huron, the eastern shore of Lake Michigan (also at least 2 locations on the western shore in Wisconsin), the western shore of Lake Huron, and the eastern part of the southern shore of Lake Superior. T. huroniana is not found south of Ludington St. Pk. in Mason County on the eastern shore of Lake Michigan or south of Au Sable Pt. in Iosco County on the western shore of Lake Huron. The westernmost locality on the southern shore of Lake Superior is the dunes at Grand Marais in Pictured Rocks National Lakeshore. T. huroniana is also present on several islands in northern Lake Michigan, including North (historical) and South Fox Islands, Hog Island (historical), High Island, Beaver Island and likely North and South Manitou Islands (habitat is appropriate and park service rangers report dune locusts). The locust is absent from the Green Bay section of Lake Michigan, both the east and west shores, even though suitable, relatively undisturbed habitat, is present (Fig. 1).

All survey localities were chosen because they had some habitat that could be classified as dune. These habitats ranged from small, single dunes to large multiple-dune complexes that extend for many miles. The dunes included relatively pristine habitats with few invasives through severely disturbed dunes with vegetation more similar to an old field. The size of the dune system (length and/or depth) was not correlated with the presence or absence of the locust. Dunes systems with and without locusts were found at both ends of the size continuum (Table 1)Table 1.Table 1..

The presence or absence of the locust was also not correlated with the presence or absence of native plant species (Table 1)Table 1.Table 1.. Particular dune plant species are just as likely to be present at sites where the locust is not found as those where it is. The most commonly noted plant species upon which the locust is known to feed (pers. obs.) were Ammophila breviligulata (present at all surveyed sites but one), Calamovilfa longifolia (at nearly all sites except those on Lake Superior), and Artemisia campestris (at most sites). Cirsium pitcheri occurred at a smaller percentage of the sites. Since an exhaustive search of habitats was not done due to time constraints, some species are, in fact, present at sites where they were not noted.

Within the known range of the locust, the occurrence of non-native, invasive plant species is somewhat correlated with locust presence or absence. Table 1Table 1.Table 1. indicates that most sites have invasives present (most commonly Centauria maculosa, Asclepias syriaca, Gypsophila spp. and old-field grasses), but only at high levels of disturbance are locust populations strongly affected. Horseshoe Bay in Mackinac County and Alcona Rd in Alcona County, are good examples of a high level of disturbance. At these sites, vegetation resembles an old field with many species of invasive plants and sand, the sand mainly stabilized. On some of these highly disturbed sites, Spharagemon collare, a banded-wing grasshopper commonly found in old fields, has become established, in 2 cases co-occurring with T. huroniana (Alcona Rd in Alcona Co. and Acme Twp Pk in Grand Traverse Co.).

Although state parks contain some of the largest and most stable populations, several sites without locusts (inside its known range) have been heavily developed for recreation within state parks (e.g., Tawas St. Pk in Iosco Co.). Some parks have extensive beachfront housing (e.g., near the city beach in Traverse City, Grand Traverse Co. and just north of Bay City St. Pk in Bay Co.)(Table 1)Table 1.Table 1.. In these cases, dunes are effectively eliminated by development, leaving no suitable habitat for T. huroniana.

The locust is replaced along the southern part of the Lake Michigan shoreline by its congener, T. maritima. This replacement is fairly abrupt, with only one known interdigitation of populations of the 2 species. T. maritima occurs at the 5^th Ave. beach in Manistee (Manistee Co.), with T. huroniana to the south at several localities in Mason Co. (Fig. 1). Further south, all populations are of T. maritima. A similar replacement does not occur along the western shoreline of Lake Huron. South of the range of T. huroniana, few suitable habitats exist until the thumb region of the state is reached (Fig. 1). Here, several T. huroniana populations reported by Hubbell (1929) were noted by Otte (1984) as having been replaced by T. maritima populations. When we rechecked these same localities in 1998, no populations of either species were located. At both Sand Pt. (where housing has eliminated nearly all dune habitat) and at Port Crescent State Park, the only banded-wing locust present was Spharagemon collare (Scudder)(Fig. 1). At Port Crescent St. Pk an extensive revegetation effort has been undertaken to re-establish typical dune grasses on previously highly disturbed dunes.

Off-road vehicle (ORV) use by itself does not appear to cause extirpation of the locust, but may significantly reduce populations. Population estimates were not made, but most sites that are heavily impacted by ORVs (now confined to the Upper Peninsula) seemed to have relatively smaller population sizes.

Species in the genus Melanoplus (not determined to species) co-occur with T. huroniana at 38 out of 45 sites where other grass-hoppers were noted, and it probably occurs at all sites (see Table 1Table 1.Table 1. for data on co-occuring species). The most common banded-wing locust occurring with T. huroniana is Dissosteira carolina, found at approximately half the sites. S. collare (at sites with some disturbance), Trimerotropis verruculata (Kirby), Camnula pellucida (Scudder) and Arphia sulphurea (F.), all occur on dunes with T. huroniana at a limited number of sites. During the early part of the summer, Chortophaga viridifasciata (DeGeer) also occurs at some of these sites. It is known to occur at Grass Bay on Lake Huron and at Wilderness State Park on Lake Michigan, at sites where the locust was monitored over the course of an entire summer. Adults of this species were not present during our mid-and late-summer surveys of the dune habitats.

Discussion

A species' distribution is defined by sites where the species is present, but the characteristics of potential habitat sites where the species is not present are key in determining the importance of the various factors for the overall species distribution. It is crucial to examine not just occupied sites, but also potential, unoccupied sites to determine the factors affecting the current distribution.

For the Lake Huron locust, whether or not a dune site is occupied is determined by a combination of historical and recent factors. The historical factors include the rise and fall of lakes during the previous several thousand years and probable competition with a congener. The recent factors are mostly anthropogenic in origin, including development and a variety of human disturbances; but the current rise and fall of lake levels may also play a significant role in the distribution pattern of the locust.

Historical factors

The historical factors that influence the distribution of the locust have their greatest influence at the periphery of its range. Past biological interactions among species are commonly considered (and will be here) when examining species distributions (e.g., insect competition studies). Geological factors are less often considered for animals which are thought to be quite mobile. However, these factors are likely to be very important in areas such as the Great Lakes with a complex glacial history (Dorr & Eschman 1970). The movement of glaciers and the resulting drastic changes in water levels in the Great Lakes have left a historical legacy in the habitats of the area, and in the distributions of plants and animals.

The original segregate population of T. huroniana [from its western relative T. pallidipennis (Burm.); see Otte 1984] was probably isolated by the movement of glaciers and then stranded on a limited number of suitable dune habitats. This same pattern of isolation from western relatives is seen in other Great Lakes species, including 2 endangered species, Hungerford's crawling water beetle (Brychius hungerfordi Spangler) and Pitcher's thistle (Cirsium pitcheri) (Voss 1996). These species with western affinities have often adapted to dry, dune areas most similar to the arid western Great Basin. The time frame for this isolation is on the order of 10,000 y (Dorr & Eschman 1970).

The locust was presumably first isolated on dunes at the edges of the ancient Great Lakes. Over the next several thousand years lake levels rose and fell dramatically, shifting the location of the coastal dune habitat (Dorr & Eschman 1970). The locust certainly moved along with the dunes, sometimes being extirpated if dunes disappeared, and other times colonizing newly forming dunes or previously unavailable ones. The absence of the locust from the dunes on the shores of the western arm of Lake Michigan (no populations found despite extensive search), including Green Bay, is a striking remnant of historical changes in lake levels. The locust occurs consistently along the shoreline of Lake Michigan from the Straits of Mackinac up to the rocky sections on the eastern shore of the Garden Peninsula in Michigan. It is then absent from all dune habitats on the western shore of the Garden Peninsula and both shores of Green Bay, even though the habitats are apparently suitable for the locust. Populations reappear again on the eastern shore of the Door Peninsula in Wisconsin (Fig. 1).

Glacial history provides a likely explanation for this absence. About 9500 y ago, when a northern outlet from the lakes opened as the glaciers receded, lake levels dropped more than 350 ft below current lake levels, drying the western arm of Lake Michigan (named Lake Chippewa at that stage) and leaving no coastal dunes in this region. Any existing locust populations would have been extirpated during this period. Subsequently, this northern outlet closed, following glacial rebound, and lake levels rose once again, refilling the western basin and re-establishing coastal dune habitats (Dorr & Eschman 1970). Although this occurred at least 4000 y ago, locusts have not recolonized these dunes, perhaps because of low vagility of the locust and unsuitable rocky habitat along the Garden and Door Peninsulas separating the existing populations from the re-established dunes.

Competition with T. maritima has left a clear pattern in T. huroniana's distribution, although the biological basis of this competition is undocumented. No locations have been found where the 2 species co-occur, even though their distributions interdigitate along the eastern Lake Michigan shoreline. Possible causes of this competition that should be investigated include interference during mating, access to food resources, or the ability to survive severe winter temperatures.

In undisturbed habitat along the southern edge of the range of the Lake Huron locust, its congener T. maritima replaces T. huroniana on lakeshore dunes as noted by Hubbell (1929) and Otte (1984). This occurs on the eastern shore of Lake Michigan in Mason Co. The cutoff is not a perfect dividing line as had been assumed by the earlier workers. At least one population of maritima (at Manistee) occurs north of some huroniana populations. Because this population is easily accessible and early surveys were spotty, this may explain the apparent discrepancy between Hubbell (1929) and Otte (1984) in their determination of the southern limits of huroniana. At the moment there is no conclusive evidence that the range of huroniana has been significantly expanding or contracting along this lakeshore.

A similar replacement of huroniana by maritima once occurred on the western shore of Lake Huron. Hubbell (1929) noted huroniana in the thumb region of Michigan (Huron Co.) and maritima farther south. Otte (1984) noted that by the 1960's maritima had replaced huroniana in the thumb region. Our current surveys indicate that neither species is present in appropriate habitats in the thumb region, but huroniana populations start again at about Au Sable Pt and continue north from there. We located no maritima populations on the eastern shore of Lake Huron, but did not examine dune systems south of the thumb region. The presumed reasons for these absences will be discussed below. Extensive surveys have not been done to determine if these 2 species show a similar pattern on the western shore of Lake Michigan in Wisconsin or on the eastern shore of Lake Huron in Ontario.

Recent factors

Our surveys indicate that anthropogenic disturbances have significantly affected populations of the Lake Huron locust. Where dunes remain, these disturbances are generally in the form of heavy human use by foot traffic and ORVs, which encourages invasive plant species to become established on the dunes. As this proceeds the habitat ceases to resemble a dune and appears much more like an old field, a habitat not used by the locust.

In the most heavily disturbed habitats, huroniana is replaced by Spharagemon collare, a banded-wing grasshopper that is common in many sandy, disturbed habitats throughout the state. This replacement may be due to either competition from S. collare or because of the habitat change leading to the replacement.

Our evidence argues for the second of these possibilities. S. collare does not replace T. huroniana, except on sites where heavy disturbance has occurred, even though it occurs on inland sites with dune-like characteristics (Otte 1984). Also, some sites exist (e.g., Horseshoe Bay in Mackinac Co.) where huroniana has disappeared, but collare has not invaded. This argues that habitat change is the key factor in the disappearance of the locust. S. collare simply invades a familiar habitat after it is established due to disturbance.

More problematic is explaining the presence of S. collare on revegetated dunes (e.g., Port Crescent St. Pk and Presque Isle Harbor Beach). These dunes have the appearance of normal T. huroniana habitat, but lack typical dune locusts. It is likely that past levels of disturbance are the key to understanding these absences. Dune revegetation is likely to occur only after dunes have been significantly affected by disturbance. This is likely to have led to enough habitat degradation that T. huroniana was eliminated from the site and S. collare became established. After revegetation of the dune, T. huroniana has not reinvaded, possibly because no source population exists (particularly true for Port Crescent St Pk) and S. collare has remained. Alternatively, the revegetation itself may have a negative impact on the Lake Huron locust. It would be very instructive to monitor future revegetation efforts as they occur, to determine their effect on T. huroniana populations.

As illustrated by the replacement of huroniana by S. collare, nearly all recent factors are associated with human disturbances. Of these, development of lakeshore dunes has resulted in much loss of habitat (e.g., much of the western shore of Lake Huron is now beachfront homes). Development has also eliminated dunes along some stretches of Lake Michigan in the Lower and Upper Peninsulas, particularly near towns or cities. For example, Hubbell (1929) noted populations of T. huroniana on the dunes near Traverse City. These populations are now extirpated by growth of the city. Similarly, new development east of Manistique in Schoolcraft Co. is likely to affect a population of T. huroniana that now exists there.

Some developments have maintained significant dune habitat and the locust remains (e.g., 7-mile Pt In Emmet Co. or Cathead Bay Drive in Leelanau Co.). The key difference seems to be maintaining not just the beach, but also a significant stretch of dune in front of or behind the housing. These examples illustrate the tolerance of the locust for moderate disturbance, along with its requirement for typical dune habitat.

Probably the most important non-anthropogenic, recent factor affecting Lake Huron locust populations is change in lake levels. At the time of our surveys, lake levels were at or above their historical average. Data from NOAA (2005) indicates that Lakes Michigan, Huron and Superior were all above the long-term average at least through 1998 (when the majority of this survey was completed). Recording stations at Mackinac City, De Tour Village, Ludington and Harrisville all indicate that the levels of Lakes Michigan and Huron were between 0.2 and 0.5 m above the long-term average of 176.6 m above IGLD 1985. These higher lake levels eliminated some potential dune habitats (particularly those linking larger dune areas), and severely reduced others. Following 1998, lake levels fell below these long-term averages.

At Grass Bay in Cheboygan Co., locust population size was monitored over a period of 8 y. Ballard (1991b) first estimated the population size at 4000 to 6000. Subsequently, the summer Biology of Insects class at the University of Michigan Biological Station estimated the population yearly from 1996 to 2000. Our first estimates in 1996 and 1997 were about 1500 individuals. The estimate decreased to 750 by 1999 and then increased again to 1300 in 2000. These population changes track changes in lake levels with a slight lag. Lake levels were 0.2 to 0.3 m below the long-term average just before Ballard's estimate. They then rose through the middle 1990's and fell again in the late 90's. Although not conclusive, this suggests that larger (and more) populations may exist when lake levels fall, while populations shrink (or become extirpated) when lake levels rise.

Understanding the links between species distributions and historical and recent factors is crucial to understanding the biology of a species, but particularly important for threatened and endangered species such as T. huroniana. The unique biology of this species limits it to a very narrow, coastal dune habitat in the northern Great Lakes. This range is further restricted by competition, the geological history of the area, and anthropogenic disturbances. This species historically had a restricted range, which has been further limited by human disturbances. Presently, this species is secure in Michigan (60+ known populations), but rare in Wisconsin and Ontario. Because of the limited and changing nature of its chosen dune habitat, it is appropriately listed as threatened in Michigan and Wisconsin.

The most important steps we could take to positively affect the conservation status of this species would be to limit lakeshore development and destruction of dune habitat. In the case of the Lake Huron locust, this does not mean a total restriction of use of the dunes. The locust appears to be quite tolerant of moderate disturbance, and may in fact depend on some disturbance to maintain its habitat. Only in cases where dunes have been effectively destroyed, either through development or overuse, has the locust been extirpated. Currently the locust is secure over most of its range.