Cliff Top Research Articles

Emplacement and movement of boulders by known storm waves — Field evidence from the Okinawa Islands, Japan

Field observations from previous studies of boulder movements were compiled to augment our own work on boulders that have been displaced by storm waves during recent typhoon events in the Okinawa Islands, Japan to elucidate the ability of storm waves to move boulders. Our observations reveal that recent storm waves displaced by sliding and overturning 100-ton boulders emplacing them on the reef or high cliff tops and that storm waves might also be capable of displacing 200 ton boulders on the reef. The weights of storm wave boulders at the Okinawa Islands are of comparable order to those boulders displaced by historical tsunami origins. Consequently, boulder weight alone is an inappropriate parameter to discriminate between tsunami or storm wave processes. However, these heavy storm wave boulders are close to the reef and cliff edges, while tsunami boulders can be deposited much further inland. Hence, horizontal displacement distance of boulders could be a useful parameter to discriminate boulders deposited by the tsunami and storm waves on the wide fringing reef. The storm wave boulders were characteristically concentrated on the southeastern (Pacific Ocean) sides of each island but large boulders are rarely found on the northwestern (East China Sea) side. This is probably because the storm wave intensities are generally stronger at the southeastern side than at the northwestern side, although differences of reef strength and initial condition of boulders should also be taken into account. Consideration of the high frequency of typhoons at the area suggests that effects of the storm waves and the consequent displacement of boulders on the reef might have contributed to the formation of the reef-moat framework that typifies the Okinawa Islands especially, if the moat is located within the transport limit of the storm wave boulders (approximately 300 m from the reef edge at the islands).

Wave-Emplaced Coarse Debris and Megaclasts in Ireland and Scotland: Boulder Transport in a High-Energy Littoral Environment: A Discussion

Scheffers et al. (2009) describe a reconnaissance survey of boulder accumulations along parts of the Atlantic coasts of the British Isles. A substantial part of their report concerns the characteristics of wave-transported boulder deposits found high above sea level on cliffs. These boulder deposits reach perhaps their ultimate development along the coasts of the Aran Islands, Galway Bay, western Ireland, and Eshaness, Shetland, northern Scotland, where evidence of wave action reaches elevations of 50 m above sea level (a.s.l.). We have previously investigated in detail these and numerous other similar localities and concluded that the boulder deposits are cliff-top storm deposits (CTSDs), part of an assemblage of features found on certain cliffs on these very high-energy coasts that relate to the impacts of extreme waves during major storms (Williams and Hall 2004; Hall et al. 2006, 2008; Hansom et al. 2008; Hansom and Hall 2009). However, Scheffers et al. (2009, p. 571) claim that at these and other sites “no evidence ... exists for storm wave dislocation of ... boulders found at altitudes of 120 m a.s.l.” Indeed, “no calculations of possible storm wave heights, storm wave physics, or ages of the deposits and no descriptions of extraordinary storms are sufficient to explain these displacements” (p. 568), and “alternative explanations, such as tsunamis, should be considered” (p. 571). Here we rebut these claims by briefly restat-

Comparison of Airborne and Terrestrial Lidar Estimates of Seacliff Erosion in Southern California

Seacliff changes evaluated using both terrestrial and airborne lidar are compared along a 400 m length of coast in Del Mar, California. The many large slides occurring during the rainy, six-month study period (September 2004 to April 2005) were captured by both systems, and the alongshore variation of cliff face volume changes estimated with the airborne and terrestrial systems are strongly correlated (r 2 = 0.95). How- ever, relatively small changes in the cliff face are reliably detected only with the more accurate terrestrial lidar, and the total eroded volume estimated with the terrestrial system was 30 percent larger than the corresponding airborne estimate. Although relatively small cliff changes are not detected, the airborne system can rapidly survey long cliff lengths and provides coverage on the cliff top and beach at the cliff base.

Rates of slope decline, talus growth and cliff retreat along the shomyo river in central japan: a space‐time substitution approach

.On the basis of a space–time substitution, rates of cliff retreat and talus development overthe pasttens ofthousands of years were quantified for valley‐side slopes along the Shomyo River with a known rate of waterfall recession. Detailed profiles of the valley‐side slopes were obtained at 19 sites along the river by map reading and field survey, and the morphologic characteristics of the slopes were then measured. By combining the locations of the slopes with the recession rate of Shomyo Falls, it is estimated that the growth rate of the talus slope is 3.9–7.2 mm/yr, the rate of decline of the cliff is 2.0–4.0 × 10−4 deg/yr, and the retreat rate of the cliff top is 6.2–11.6 mm/yr.

Comparison of short-term seacliff retreat measurement methods in Del Mar, California

Seacliff retreat has been variously characterized as the recession rate of the cliff top, of the cliff base, and as the bulk recession rate based on volumetric changes of the entire cliff face. Here, these measures of retreat are compared using nine semi-annual airborne LiDAR (Light Detection And Ranging) surveys of southern California seacliffs. Changes in the cliff base location (where the steeply sloping cliff face intersects the beach) include cliff retreat owing to basal erosion, but also reflect changes in beach sand level and basal talus deposits. Averaged over the 2.5 km alongshore study span, the cliff base actually prograded seaward about 12 cm during the 4-year study. Cliff top change was dominated by few, relatively large (several meters) localized retreats. Cliff face changes, that include failures and deposits anywhere on the cliff profile, had a relatively small mean magnitude compared to cliff top changes and were more widely distributed alongshore. However, the similar alongshore averaged, cumulative cliff top and net bulk cliff face end-point retreat (14 and 19 cm, respectively) suggest that mean cumulative cliff top retreat can potentially be a viable surrogate for mean net cumulative cliff-wide erosion (and vice versa) over relatively short time periods. Cliff face erosion occurred repeatedly at some locations, confirming the presence of seacliff erosion hot-spots during the study period.

Composition, Phytogeography, and Relict Status of the Vascular Flora of Alvars and Cliff Tops Southwest of Great Slave Lake, Northwest Territories, Canada

Alvars in the vicinity of Enterprise, Northwest Territories, had 87 species of vascular plants. Of these, 14 were frequent or at least locally abundant, including: Juniperus horizontalis, J. communis var. depressa, Geum triflorum var. triflorum, Carex richardsonii, Populus tremuloides, Elymus trachycaulus subsp. trachycaulus, Koeleria macrantha, Saxifraga tricuspidata, Artemesia campestris subsp. borealis, Senecio lugens, Arctostaphylos uva-ursi, and Galium boreale. Of the trees comprising the surrounding forest, only Populus tremuloides and Pinus banksiana were common in the open areas. The cliff tops had fewer species but they also had some species of arctic and high boreal affinity that were absent from the alvars. Associations characterized by Deschampsia caespitosa occurred in lower areas of alvar that collect run-off, whereas higher areas were characterized by a variety of plants including Carex richardsonii, Elymus trachycaulus, and Geum triflorum. Cracks in the limestone pavement in elevated areas were occupied by Pinus banksiana and Juniperus communis. Approximately 12% of the species present were rare and/or restricted, including Avenula hookeri, Carex filifolia, and Dodecatheon pulchellum. Compared to the alvars of the Great Lakes region, the alvars and cliff tops of Northwest Territories had a much higher percentage of species that were also found in the Beringian region of northwestern North America. Higher numbers of boreal, arctic, and cordilleran species also contributed to the distinctive nature of Northwest Territories alvars, based on clustering of similarity coefficients from species presence or absence data. The alvar flora of Northwest Territories is believed to have been established in the region of Great Slave Lake between 10,000 and 8000 years ago, prior to invasion of the boreal forest. At this time, the northwestern Beringian and Cordilleran floras met the retreating tundra and prairie floras from the south in an expanding ice-free corridor. The relict flora of Northwest Territories alvars and cliff tops includes species of the Beringian steppes and tundras such as Plantago canescens, species of the central North American plains such as Geum triflorum, and many species that could have originated from the prairie-like habitats of either region. This unusual relict flora, originating in early postglacial times, is now isolated within the present boreal forest that spread through the region approximately 8000 years ago.

Competing mechanisms for boulder deposition on the southeast Australian coast

This study investigates the role of late Holocene sea-level change, large storms and possible pre-historic tsunami in the deposition of boulder features on an exposed headland and sheltered rock ramp. Large accumulations of boulders are found on coastal rock platforms, cliff tops and ramps in the Jervis Bay region of southeastern Australia. These deposits are elevated above sea-level and in places exhibit obvious signs of imbrication as a response to flow in a landward direction. The event history of these features is controversial and Holocene sea-level change, storms and tsunami can be considered as possible, non-mutually exclusive mechanisms of deposition. It is apparent that even if the detachment site, transport distance, elevation and final orientation of a boulder can be identified, deciphering the event history remains a difficult task.

Anthropogenic modifications in the erosional rhythm of a coastal cliff. Rocha do Gronho (western coast of Portugal)

The beach-cliff system “Rocha do Gronho” is a stretch of the west coast of Portugal that marks the southern limit of a coastal lagoon – The Obidos lagoon. The cliff is cut in sedimentary material and the elongated beach is generally narrow being almost submerged by the spring tides. Existing documents show that, in almost all of the second half of the twentieth century, there has been a state of equilibrium of the beach-cliff system because the abundance of sediments proceeding from the lagoon and carried south by the longshore current (predominantly N-S) provided a covering of the cliff base that protect it from the assailing force of the waves. The cliff face evolved mainly by small magnitude rockfall events. Between 1958 and 1995, a rate of retreat of 0.15m (±0.02m) year−1 was recorded. Afterwards, the frequency and the magnitude of the occurrences of mass movements had changed. Five events leading to a 33.6m retreat of the cliff top have been recorded. The type of movement, the affected length of the top of the cliff, the maximum retreat, the lost area of the top of the cliff and the volume of material displaced were estimated for each event. The identification of the processes involved as well as the evaluation of the potential triggering factors allowed an assessment of the influence of human intervention in the lagoon (changing the main draining paths and consequently the patterns of sediment transport) in the modifications of the “Rocha do Gronho” evolution.

Magnitude-frequency of sea cliff instabilities

Abstract. The magnitude-frequency relationship of sea cliff failures in strong, low retreat rate cliffs, was studied using systematic historical inventories carried out in the coasts of Portugal and Morocco, in different geological and geomorphological settings, covering a wide size scale, from small to comparatively large rockslides, topples and rockfalls, at different time and spatial scales. The magnitude-frequency expressed in terms of volume displaced and of horizontal area lost at the cliff top showed good fit by inverse power laws of the type p=a.x−b, with a values from 0.2 to 0.3, and exponents b close to 1.0, similar to those proposed for rockfall inventories. The proposed power laws address the magnitude-frequency for sea cliff failures, which is an important component of hazard assessment, to be completed with adequate models for space and time hazard components. Maximum local retreat at the cliff top provided acceptable fitting to inverse power laws only for failures wider than 2m, with a = 4.0, and exponent b = 2.3, which may be useful to assess the cliff retreat hazard for the use of areas located near the cliff top.

Patterns and rates of erosion produced by high energy wave processes on hard rock headlands: The Grind of the Navir, Shetland, Scotland

The Grind of the Navir is an ignimbrite headland on the exposed Atlantic coast of the Shetland Islands, Scotland. During storms, offshore wave heights exceed 20 m and deep water close inshore allows high-energy waves to impact on these cliffs. The cliff top at ∼ 15 m above sea level is awash with wave water when wave heights exceed 8 m, a condition met in storms in most years. Detailed mapping, ground photography and patterns of lichen colonisation and growth allow the processes, patterns and rates of erosion to be assessed on different parts of the headland over the past 100 years. Recent fractures in the rock of the cliff face and top indicate that wave impact forces exceed the 1.5 kPa tensile strength of the rock and fresh sockets low on the cliff face record the removal of fracture-bounded blocks that are lost to seaward. On the upper part of the cliff face, upward moving water flow is capable of removing blocks from both vertical faces and stepped overhangs. Clefts in the cliff face are extended inwards and upwards by crack propagation and block removal, leading to the development of slot caves. The movement of waves across the cliff top in fast-moving bores quarries blocks from the cliff face close to the edge and from rock steps on the cliff top, promoting further rock fracturing. Socket sizes and block characteristics indicate that blocks of > 1 m 3 are rotated from sockets on the cliff top platform and carried landwards for up to 60 m to be deposited in a series of boulder ridges on the cliff top. Multiple boulder movements during a storm in January 2005 generated impact marks and orientated striations that allow the pattern of water movement to be reconstructed across considerable areas of cliff top. The average rate of erosion at The Grind is estimated at 1.3–6.0 mm/yr on the cliff face and 5 mm/yr on the cliff top, although erosion rates vary in both time and space. Although block movement on the cliff top occurs in most years, erosion of rock by block removal is most active during major storms. Nevertheless, some parts of the cliff top and face remain blackened by lichen and so have experienced little recent erosion over the last 70 years or so. Although erosion occurs at the base of the cliffs, in recent decades it has been concentrated on upper part of the cliff face and top. Cliff-top erosion is most intense where stepped geos act as channels for waves to access the cliff-top. The most active features on the cliff top are the boulder ridges where substantial changes can be traced to storms over the past century. Unlike conventional models of cliff erosion where erosion is concentrated by wave attack at the waterline, wave impacts on this headland may occur on any part of cliff face and top and appear particularly effective on the upper cliff and cliff top platform.

Cliff retreat and talus development at the caldera wall of Mount St. Helens: Computer simulation using a mathematical model

To simulate the landform evolution at the caldera wall of Mount St. Helens, USA, a mathematical model for talus development was applied to model the topographic change during the 11years from the volcanic eruption, i.e., from formation of the cliff. Simulated results show that the topographic change is predicted to be large for about 10years after the eruption and to decline thereafter. If snow accumulation in the talus slope deposits is negligible, the talus top will not reach the cliff top within 300years after the eruption. Talus growth in Mount St. Helens was much faster than that in the Chichibu Basin, Japan. This may indicate the low strength and/or high weathering rate of the rockwall of Mount St. Helens, resulting in rapid production of debris and rapid retreat of the cliff.

Short and medium-term evolution of a coastal sector in Cadiz, SW Spain

The present paper is a combined study on the medium and short-term evolution of the littoral between Chipiona and Rota (SW Spain). The analysis of coastal evolution over a period of 24 years (i.e. medium-term), was carried out using six, well temporally spaced, photogrammetric flights. Cliff top and dune toe were used as shoreline indicators to solve problems related to the use of watermark in tidal environments. Short-term littoral variations were monitored monthly by means of an electronic theodolite over a period of 2 years. The width variations of the dry beach were surveyed by always using a constant, average value for the mean sea level position derived from the numerous surveys carried out. The obtained data was representative of both seasonal and annual variations. Over medium-term, most of the littoral recorded erosion, while accretion has been recorded over recent years in the southern part. Coastal erosion was related to the impact of several storms and dune accretion was linked to the action of eastern winds. Over the short-term time frame, about the same percentage of coastal erosion and accretion was recorded. Most of the important accretion trends were observed at the central and southern parts of the littoral, the largest amount of erosion being recorded in the southern end of the littoral, in a nourished beach. All the studied beaches (except Punta Candor dunes) showed more short-term changes than medium-term ones, because of different factors, essentially the great seasonal variability of the studied beaches. The partial discrepancy between the medium and the short-term trends was related to the applied methodology, which demonstrates the problem of comparing different coastal features (dunes and cliffs on one hand, and beaches on the other) and to the implications of 2–3 years of lasting fair weather conditions that favoured beach and dune accretion, within an historical retreat tendency related to storm actions.

Distribution, geomorphology and lithofacies of cliff-top storm deposits: Examples from the high-energy coasts of Scotland and Ireland

Cliff-top storm deposits (CTSDs) occur on cliffs at elevations of up to 50 m above sea level at exposed sites on the deep-water coasts of the British Isles. This study examines the distribution, geomorphology and lithofacies of CTSDs at sites from Shetland, Orkney, Caithness and the Outer Hebrides in Scotland and from the Aran Islands in Galway Bay in Ireland. CTSDs are generated largely by the quarrying of blocks from the cliff top, and transported by green water bores across cliff-top platforms and ramps to be deposited in backing ridges or as debris spreads. Maximum boulder sizes reach 48 m 3 but it is likely that much larger blocks can be quarried, prior to disintegration during transport. Eye-witness accounts and field mapping demonstrate that formation and modification of CTSDs has continued during major storms over recent decades. Recent CTSDs bury a range of man-made debris but older deposits lack this and instead show weathering effects that indicate a longer residence time. In Shetland OSL dates on intercalated sands suggest that the oldest CTSDs date from ∼ 800 AD. Radiocarbon dates on shell buried inside CTSD ridges relate to major storms between 1700 and 1900. CTSDs represent an overlooked archive of storm sedimentation that has great potential for the elucidation of local storm chronologies. CTSDs also provide evidence of wave impacts on any part of the cliff face or top, in fundamental contrast to the concentration of wave action at the cliff foot implicit in traditional models of the erosion of rock coasts.

Genetic structure and gene flow in wild beet populations: the potential influence of habitat on transgene spread and risk assessment

Summary The consequences of the movement of transgenes from genetically modified (GM) crops into wild populations of plants continues to be of concern to ecologists and conservationists because of the possible threat posed to those populations in terms of their continued survival and because of the further knock‐on effects that might occur to habitats in which they occur. We examined five UK sea beet Beta vulgaris ssp. maritima populations from each of two major habitat types, cliff top and drift line. We assessed population genetic parameters, genetic diversity, gene flow, population differentiation and isolation by distance, to enable determination of the likelihood and consequences of spread to wild populations of genes from cultivated sugar beet group Beta vulgaris ssp. vulgaris, which could in the future be transgenic. Drift line populations were more diverse than cliff top populations and also showed greater levels of gene flow. Isolation by distance was identified in both habitats, but the relationship between genetic and geographical distance was detectable over longer distances for drift line populations. However, clear indications of vicariance (the subdivision of a population into distinct taxa by the appearance of a geological barrier) between cliff and drift line populations were also evident, because of the restriction of gene flow between the two habitats occurring more in one direction than the other. Synthesis and applications. The likelihood of transgene spread from crop to wild populations is habitat dependent and conservation management decisions could therefore vary from one population to another, for example water courses were found to facilitate seed dispersal. This should be taken into account when estimating isolation distances for GM beet, and when predicting transgene frequencies (exposure estimates) for environmental risk assessments of GM beet.

Influence of Habitat Heterogeneity on Distribution, Occupancy Patterns, and Productivity of Breeding Peregrine Falcons in Central West Greenland

Abstract We used occupancy and productivity data collected at 67 cliffs used for nesting from 1972 to 1999 to assess patterns of distribution and nest-site selection in an increasing population of Peregrine Falcons (Falco peregrinus) in central West Greenland. Peregrine Falcons breeding at traditionally occupied cliffs used for nesting had significantly lower variation in productivity and thus these cliffs were better quality sites. This indicates that Peregrine Falcons occupied cliffs according to a pattern of despotic distribution. Falcons breeding at cliffs that were consistently occupied during the breeding season had higher average productivity and lower variation in productivity than falcons at inconsistently occupied cliffs, and thus consistent occupancy also was indicative of cliff quality. Features of high quality habitat included tall cliffs, greater change in elevation from the lowest point within 3 km of the cliff to the cliff top (elevation gain), and protection from weather on the eyrie ledge. Spacing of suitable and occupied cliffs also was an important feature, and the best cliffs generally were more isolated. Increased spacing was likely a mechanism for reducing intraspecific competition. Our results suggest that Peregrine Falcons use a resource defense strategy to compete for better quality habitats and may use spacing and physical features of a nest site to identify good quality breeding habitat.

INFLUENCE OF HABITAT HETEROGENEITY ON DISTRIBUTION, OCCUPANCY PATTERNS, AND PRODUCTIVITY OF BREEDING PEREGRINE FALCONS IN CENTRAL WEST GREENLAND

Abstract We used occupancy and productivity data collected at 67 cliffs used for nesting from 1972 to 1999 to assess patterns of distribution and nest-site selection in an increasing population of Peregrine Falcons (Falco peregrinus) in central West Greenland. Peregrine Falcons breeding at traditionally occupied cliffs used for nesting had significantly lower variation in productivity and thus these cliffs were better quality sites. This indicates that Peregrine Falcons occupied cliffs according to a pattern of despotic distribution. Falcons breeding at cliffs that were consistently occupied during the breeding season had higher average productivity and lower variation in productivity than falcons at inconsistently occupied cliffs, and thus consistent occupancy also was indicative of cliff quality. Features of high quality habitat included tall cliffs, greater change in elevation from the lowest point within 3 km of the cliff to the cliff top (elevation gain), and protection from weather on the eyrie ledg...

Nesting and Disturbance of the Black-faced Spoonbill in Liaoning Province, China

The Black-faced Spoonbill (Platalea minor), breeding on Xing-Ren Tuo, arrived from late March to April and left from mid-July to August. The summer group included breeding pairs, non-breeding adults and sub-adults, while some transient individuals were observed irregularly. After arrival, the birds slept and rested on bare rock on the seashore or cliff top. They departed before dusk to feed, returning in the early morning. Pairs formed 15 days before the first egg was laid. The copulation frequency averaged 0.38±0.18 times per hour after pairs formed, and the highest frequency occurred in the early morning. The clutch size was two or three eggs, and the incubation period lasted 26.2±1.5 days (range 25-29, N = 6). Juveniles left their nest after 30 days and could fly freely after 36 days. From 1999 to 2003, the hatching rate of 29 eggs was 86% and the survival rate to fledging of nestlings amounted to 84% of 25 hatched chicks.

Terrestrial laser scanning for monitoring the process of hard rock coastal cliff erosion

Hard rock cliffs represent approximately 75% of the world's coastline. The rate and nature of the mechanisms that govern the retreat of these cliffs remain poorly constrained, primarily because conventional approaches employed to monitor these processes are generally inadequate for describing cliff erosion processes directly. These techniques are usually centred upon the interpretation of data collected periodically from aerial sensors, including stereographic aerial photographs and more recently air-borne LIDAR. These methods are generally not capable of assessing the pattern of erosion on the cliff face due to the oblique viewing angles, and hence tend to concentrate upon the resultant recession of the cliff top rather than change on the cliff face. Thus, processes of undercutting and small scale iterative failures of localized sections of the cliff face are generally not recorded. It is only when a failure affects the cliff top that any retreat is recorded. It is therefore unsurprising that cliff erosion is commonly deemed to be episodic. This paper presents a new approach to detailed cliff process monitoring using terrestrial laser scanning (TLS), which directly monitors changes on coastal cliff faces. The method allows the quantification of failures ranging in scale from the detachment of blocks of a few centimetres in dimension through to large rock, debris or soil, falls, slides and flows over 1000 m 3 . The collection of data is on-site and rapid and hence cost effective, providing a detailed description of the nature of coastal cliff erosion. This paper describes the methodological approach and demonstrates the range of results which can be generated, here shown for 16 months of monitoring data collected for a near-vertical cliff section on the coast of North Yorkshire, UK. The results demonstrate that terrestrial laser scanning can be used to quantify cliff failures to a previously unobtainable precision. The results reveal a strong spatial and temporal pattern of cliff collapse which contradicts commonly held perceptions of the nature of coastal cliff development.

Estimating long-term cliff recession rates from shore platform widths

Coastal cliff erosion is a problem in many coastal areas. However, often only very limited data are available to quantify the rates of recession for the development of coastal management strategies. In the soft flysch deposits of the Waitemata Group, Auckland, New Zealand, coastal cliffs are associated with shore platforms. Two models exist for the profile evolution of shore platforms and associated cliffs: the first suggests that an equilibrium profile develops in response to erosive processes, and this profile subsequently migrates landward; the second model suggests that the seaward margin of the shore platform is relatively static, and the profile extends landward through a combination of cliff recession and platform lowering. Physical simulations and field measurements for mudstone and limestone lithologies indicate that the second model is more likely for soft flysch deposits. A eustatic sea-level curve for the Weiti Estuary, Auckland, suggests that up to 7120 ± 70 years have been available for shore platform development since sea level reached the present seaward margins of shore platforms. Shore platform widths were measured using GPS at two sites in Waitemata Group rocks: the North Shore of Auckland; and the southern side of the Tawharanui Peninsula, North Auckland. The long-term cliff recession rates estimated from shore platform widths (1.4 ± 0.1 to 14.3 ± 0.1 mm y − 1 ) are consistent with the lower end of the average range of cliff top and face recession rates published for Waitemata Group rocks using different methods (11–75 mm y − 1 ), and in agreement with cliff base recession estimates (∼3.5 mm y − 1 ). Shore platform widths were qualitatively related to the rock mass characteristics of the associated cliffs, and therefore platform widths could provide a method of identifying regions of potential hazard.

CANOPY STRUCTURE, MICROCLIMATE, AND HABITAT SELECTION BY A NOCTURNAL SNAKE, HOPLOCEPHALUS BUNGAROIDES

Forest structure strongly influences ambient environmental conditions such as light and temperature, but most studies on habitat selection by mobile organisms have either ignored canopy structure or treated it as a dichotomous variable (e.g., shady or sunny). Furthermore, the predominance of active diurnal species as model organisms in such studies has left many unanswered questions about the importance of vegetation-related variables for nocturnal and sedentary species (e.g., what does shade mean to an organism that moves at night and sits in a cave all day?). We used hemispherical photography to quantify canopy structure and examine its role in determining the thermal microenviron- ments available to a rock-dwelling nocturnal snake (Hoplocephalus bungaroides) across two different spatial scales. The narrow plateaus inhabited by the snakes in southeastern Australia are highly heterogeneous with respect to vegetation: east-facing aspects are dense- ly covered whereas west-facing aspects are patchy mosaics. We found that temperatures of potential retreat sites increased with increasing canopy openness, but the definitive deter- minant of retreat-site temperature was incident radiation intensity, which depended upon the location of canopy gaps relative to the sun path. This factor restricted the snakes to west-facing cliff tops, and there only to an optimal subset of rocks that received adequate irradiance. Moreover, thermal regimes of retreat sites displayed higher maxima and were evening shifted relative to randomly sampled rocks in the same area. Our results suggest that thermally suitable retreat sites are a limiting resource, and that local increases in vegetation density might contribute to the decline of this endangered species.