A complex array of widespread, but domainally developed, structures is preserved in the Tuolumne batholith, including stationary and migrating tubes, pipes, troughs, diapers, and plume heads. These structures, all formed by local magma flow through crystal-mush host magmas, are often associated with the formation of schlieren rich in accessory and mafic minerals, and are associated with filter pressing and accumulations of crystals with diverse magma histories and ages. Together they represent a network in which channelized flow occurred in an existing chamber of crystal-rich magmas, resulting in local compositional and structural diversity.

These structures also are useful structural tools for evaluating the internal evolution of magma chambers. For example, the consistently steep tube and pipe axes indicate that neither the pluton nor features in the pluton were tilted during growth, thus excluding models in which subhorizontal layers tilted to form the existing steep contacts. Although the overall direction of younging established by geochronologic studies is toward the batholith center, local younging directions determined from troughs cutoffs indicate that outward growth occurred in many zones. The highly variable movement directions of local diapirs and plumes require interactions between buoyancy forces and other gradients.

The existence and characteristics of these structures have several other implications. Interpretations derived about crystal residence times in chambers and about crystal mixing during eruptions need to be treated with caution, since mixed crystal populations existed well prior to eruptions, and in the case of the Tuolumne batholith involved crystals with ages ranging over ~10 m.y. A likely solution is that crystals in subvolcanic chambers become armored (rimmed) by other crystals or exist in crystal clusters that, in spite of changing environmental conditions, prevent rapid chemical communication with the surrounding melts. These structures also challenge many aspects of the incremental chamber growth model resulting in sheeted bodies championed by Glazner, Bartley, Coleman, and colleagues for the Tuolumne batholith. The thousands of preserved internal structures provide clear evidence against late annealing and removal of internal contacts, and are difficult to reconcile with either vertical sheeted or subhorizontal laccolith models; however, they are permissive of early pulsing leading to one or more large magma chambers.


It has been suggested that as magmas approach their solidus and become crystal rich, their viscosities rise to such a degree that it becomes essentially impossible for them to convect, fractionate, and/or erupt and thus to form the compositional and structural diversity often preserved in chambers (Barriere, 1976; Brandeis and Marsh, 1989; McBirney, 1993; Vigneresse et al., 1996; Scaillet et al., 2000; Dingwell, 2006). This has led some to conclude that most preserved internal compositional and structural variations in plutons reflect either the juxtaposition of many pulses derived from the melting of a heterogeneous lower crust or mantle, or the processes that operated early during ascent and chamber construction rather than processes operating in an already constructed magma chamber (e.g., McNulty et al., 1996; Coleman et al., 2004, 2005; Glazner et al., 2004; Gray et al., 2008). This view has been questioned by authors describing a number of processes that occurred in largely constructed chambers, such as local ponding of mafic magmas, convection, mixing, and fractionation, which occurred in crystal mush zones that approached and eventually exceeded 50% crystals (e.g., Wiebe, 1996; Wiebe and Collins, 1998; Marsh, 1996, 2006; Miller and Miller, 2002; Hersum et al., 2005; Walker et al., 2007), and is being challenged further by recent studies concluding that processes leading to compositional diversity occur in magmas with >50% crystals (e.g., Bachmann and Bergantz, 2004, 2008; Žák and Klomínský, 2007). For example, Bergantz (2000) used numerical modeling to examine rheological controls of internal magma boundaries and concluded that if an intrusive unit along an internal margin is not fairly crystal rich and thus stiff, its boundary is not stable and would collapse, making it less likely that margins between crystal poor magmas are preserved in chambers. A related issue comes from a dramatic proposal that many of these internal contacts, including internal contacts between magma pulses, are entirely removed due to late, thermally driven annealing (Glazner et al., 2008a, 2008b) a suggestion that was challenged by Vernon and Paterson (2008a, 2008b). Thus, the timing of formation of magmatic structures and associated crystal percent will dramatically influence our interpretation of chamber construction and evolution and inferences about when magmas can mix, mingle, and fractionate in chambers and their subsequent behavior during volcanic eruptions.

These issues also are critical for evaluating another recent conclusion that both volcanic and plutonic rocks are mixtures of crystals with distinct histories. For example, Davidson et al. (2001, 2005, 2007) and others (Broxton et al., 1989; Christensen et al., 1995; Claiborne et al., 2006; Cooper and Reid, 2003; Costa et al., 2003; Barbey et al., 2008; Ramos and Reid, 2005; Wallace and Bergantz, 2005; Walker et al., 2007) have demonstrated, using isotopic fingerprinting in single minerals, that crystal exchange between different melts is a common phenomenon and that the resulting crystal populations are often accumulated from two or more sources. Recent high precision U/Pb thermal ionization mass spectrometry zircon dating of multiple single grains supports this conclusion through the recognition that zircon populations in a single sample are a mix of xenocrysts, antecrysts, and autocrysts (e.g., Brown and Fletcher, 1999; Charlier et al., 2005; Bindeman et al., 2006; Claiborne et al., 2006; Matzel et al., 2006b; Miller et al., 2007; Walker et al., 2007). The resultant volcanic or plutonic rock is thus a mechanical mixture of crystals, for which (1) the bulk rock geochemistry may say little about equilibrium processes of melting and crystallization; (2) the contributing sources (mantle, crust, subsequent contamination) may be obscured; and (3) the age may be misleading or at best leaves untapped a great deal of useful information.

Another implication of the above observations is that some volcanic eruptions are derived from magma chambers constructed from numerous pulses of magma that accumulated over time and are integrated just prior to or even during eruption. This requires fairly late mixing of diverse crystal populations in magma chambers. Alternatively, could mixing of diverse crystal populations occur during ascent and magma chamber construction and then this disequilibrium assemblage be preserved until eruptions occur? For example, in the Tuolumne batholith, California, mixing of distinct crystal populations is nicely supported by recent U/Pb zircon data (of Matzel et al., 2005, 2006b, 2007; Miller et al., 2007) in which widespread zircon antecrysts with concordant ages identical to older units in the batholith are found in the younger phases, an observation increasingly established in other magmatic systems (Gardner et al., 2002; Bacon and Lowenstern, 2005; Claiborne et al., 2006; Bachmann et al., 2007). How and when did this mixing of zircons occur?

Answers to these questions depend on to what degree late processes in middle and upper crustal crystal-rich magma chambers can sort magma into compositionally different components, form magmatic structures that do not reflect source or ascent processes, and mix diverse crystal populations. Here I describe an impressive array of compositionally and texturally defined magmatic structures such as stationary and migrating tubes, pipes, troughs, small-scale within-chamber diapers, and plumes in the Tuolumne batholith, central Sierra Nevada, California. Their characteristics and relative timing indicate that their formation and preservation required development in crystal-rich (e.g., >50%) magmas, and thus indicate that late, local movement of magmas resulted in crystal accumulations (of minerals with diverse histories), fractionation, and formation of these structures (Weinberg et al., 2001; Paterson et al., 2005; Žák and Klomínský, 2007; Vernon and Paterson, 2008a; Ruprecht et al., 2008; Bachman and Bergantz, 2008).

An examination of these structures also indicates that they are useful tools for addressing the internal evolution of magma chambers (e.g., Fernandez and Gasquet, 1994; Wiebe and Collins, 1998; Pignotta et al., 2006). For example, tubes and pipes, but not plumes and diapirs, may preserve information about paleohorizontal and local flow directions. Troughs provide information about local younging (or growth) directions as well as directions of magma flow. Diapirs record information about local gradients and displacement fields. All of these magmatic structures record information about processes by which crystals and melt are sorted, sometimes separated, and other times mixed, and thus about processes resulting in compositional and structural diversity formed in middle and upper crustal magma chambers.


The Tuolumne batholith is an ~1100 km2 Late Cretaceous composite batholith exposed in the central Sierra Nevada, California (Fig. 1). It is emplaced into Cretaceous granitoids (e.g., 102 Ma El Capitan granite) and amphibolite-grade metasedimentary rocks (Kings Sequence) to the west and older plutonic and greenschist-grade metavolcanic rocks of Triassic to Cretaceous age to the east (Huber et al., 1989; Schweickert and Lahren, 1993, 2006). Contacts of the batholith with host rocks are steeply dipping and generally discordant to the host rock structures, although local (100 m) domains occur in which older structures are deflected and margin-parallel foliations and steeply plunging lineations occur. Local relief establishes that these steep contacts extend at least 2 km in height. Gravity data (Oliver, 1977; Oliver et al., 1987) indicate that many of these vertical contacts may extend downward at least to depths 6–12 km below the present erosion level for most units, although a recent gravity study concluded that the central Johnson granite porphyry may only extend down a few kilometers (Titus et al., 2005). Estimates of emplacement depth of the currently exposed surface of the Tuolumne batholith, based on Al-in-hornblende barometry (Ague and Brimhall, 1988; Webber et al., 2001; Gray, 2003; Anderson et al., 2007), indicate a depth of 6–10 km, consistent with widespread andalusite and local sillimanite in the surrounding host rocks (Rose, 1957; Memeti et al., 2005a; Anderson et al., 2007).

The Tuolumne batholith (Fig. 1) consists of four nested, progressively more evolved inward, intrusive units: (1) the outer Kuna Crest unit to the east and its equivalents along the western and southern margins (tonalites of Glen Aulin and Glacier Point, granodiorite of Grayling Lake) and inner phases, including (2) the Half Dome granodiorite, (3) the K-feldspar megacrystic Cathedral Peak granodiorite, and (3) a central phase, the Johnson granite porphyry (Bateman, 1992; Bateman and Chappell, 1979). The Kuna Crest unit is mostly fine- to medium-grained, dark colored, equigranular tonalite, quartz diorite, and biotite-hornblende granodiorite, typically with a strong magmatic fabric and abundant mafic enclaves. The Half Dome granodiorite, consisting of an outer equigranular and inner porphyritic phase, is generally much coarser grained than the Kuna Crest unit and is characterized by the presence of large (to ~2 cm in length) prismatic euhedral hornblendes and conspicuous sphene. The porphyritic variety contains large (to ~6 cm in length) K-feldspar phenocrysts. The Cathedral Peak granodiorite, which forms the most voluminous part of the currently exposed Tuolumne batholith, consists of biotite granodiorite containing abundant large K-feldspar phenocrysts (on average 4–5 cm, but locally to ~12 cm in length) and large quartz crystals in a medium-grained matrix. The central phase, the Johnson granite porphyry, is fine-grained equigranular biotite granite, locally containing sparse antecrystic K-feldspar phenocrysts and other fragments from the Cathedral Peak granodiorite. Work by Memeti et al. (2005b, 2007) indicates that all these main units show internal compositional variations, particularly in lobes extending out from the main batholith, in which the main units fractionate to form central leucogranite lenses (Fig. 1).

In contrast to the model of in situ fractionation of single parent melt proposed by Bateman and Chappell (1979), geochronological studies (Kistler and Fleck, 1994a, 1994b; Coleman and Glazner, 1997; Matzel et al., 2005, 2006b, 2006c, 2007; Miller et al., 2007; Memeti et al., 2007) indicate that the batholith was constructed over a 10 m.y. duration between 95 and 85 Ma. Based on Sr and Nd isotopic analyses, Kistler et al. (1986) concluded that the inner and outer intrusive units evolved separately and were largely derived from mantle (outer units) and crustal (inner units) derived magmas. Coleman et al. (2004) and Glazner et al. (2004), using a new geochronologic data set, concluded that the entire batholith was constructed of thousands of dikes or sills, and that these dikes did not interact significantly with neighboring pulses. In their model, numerous internal contacts formed, and little to no fractionation, mingling, convection, internal margin collapse, stoping, and related processes occurred because a large magma chamber never existed in this batholith. Coleman et al. (2008) and Glazner et al. (2008a, 2008b) recently tempered this view and concluded that a large magma chamber formed by stacked laccoliths may have existed (Coleman et al., 2008), but that widespread annealing removed evidence of earlier internal contacts.


An impressive variety of internal magmatic structures is preserved in the Tuolumne batholith. Here I focus on five types of locally developed but widely distributed structures (Fig. 2; Table 1): (1) stationary and migrating tubes (Figs. 3–6) (magma tubes have been called snail structures or ladder dikes; e.g., Reid et al., 1993; Weinberg et al., 2001); (2) pipes (Figs. 7 and 8; see also Wiebe, 1996); (3) troughs (Figs. 9 and 10); (4) diapirs (Figs. 11 and 12); and (5) plumes (Fig. 13). Table 1 and Figure 2 contrast the main differences between these structures. My research group's recent 1:24,000 mapping of ~60% of the batholith indicates that all 4 major units preserve examples of these structures, that their formation was times transgressive from after 95 Ma to ca. 85 Ma, and that they are most common in the Half Dome and Cathedral Peak granodiorites, less common in the Kuna Crest granodiorite, and rare in the Johnson granite porphyry and other leucogranite lenses. Our mapping also indicates that thousands of tubes, plumes, and troughs exist, whereas pipes and diapirs are much less common: the latter two structures are more commonly located in areas of well-developed magmatic sheeting (Žák and Paterson, 2005; Paterson et al., 2008). Troughs are common along internal contacts between the main units but also occur in zones between these contacts.

I also examine herein the possibility that there may be gradations and/or genetic connections between some of these structures (Figs. 14–18); however, I first describe compositional and structural characteristics of end-member examples.

Stationary and Migrating Tubes

Magma tubes are defined as cylindrical or tube-shaped structures in three dimensions that in sections perpendicular to tube axes, display numerous, enclosed (if not removed by subsequent magmatic erosion), elliptical schlieren bounded layers (Figs. 2 and 3; Table 1). At least some and occasionally all layers in the tubes have compositions and/or textures distinct from the surrounding host magma (Figs. 3–5). Two general types of tubes occur in the Tuolumne batholith: stationary tubes in which the tube axes do not migrate with time (Figs. 2 and 3) (see also Weinberg et al., 2001), and migrating tubes in which the tube axis moves and develops path lengths of meters to many tens of meters (Fig. 4) (see also ladder dikes of Reid et al., 1993). Particularly in the migrating tubes the displacement of tube centers result in crescent-shaped patterns of alternating light and dark schlieren along the path lengths (Fig. 4; Burgess, 2006; Burgess and Miller, 2008).

Both stationary and migrating tubes are found in all four units in the Tuolumne batholith, although they are rare in the Johnson granite porphyry and in similar leucocratic granitic lenses recently discovered in other Tuolumne batholith units by Economos et al. (2005) and Memeti et al. (2005b, 2007). Initial estimates suggest that there are thousands of tubes in the Tuolumne batholith with a roughly equal number of stationary and migrating tubes. Tube densities measured to date vary from 0 to hundreds per square kilometer, indicating that they are spatially clustered. This clustering does not appear to be related to internal contacts.

Maximum tube diameters vary from a few centimeters (Fig. 3C) to >40 m. Larger tubes may exist but may be difficult to recognize because schlieren boundaries of large tubes would appear as planar schlieren zones unless they are followed along strike. The crosscutting relationships between schlieren-bounded layers (Figs. 3–5) and mineral grading in schlieren indicate that both stationary and migrating tubes decrease their tube radii through time, often resulting in fairly small (centimeters to decimeters) final tube diameters. In stationary tubes the decrease in diameters can be either symmetrical or slightly asymmetrical, but the tubes are always nested (Fig. 3). When exposed in three dimensions, most tubes do not change their general characteristics over the short vertical distances exposed (Fig. 5).

Almost all tube axes plunge >70° (Figs. 5 and 6). Since most tube axes are steep, the maximum vertical dimensions of tubes are not well constrained: local relief indicates minimum vertical dimensions of at least several meters. In rare cases tubes get reintruded by host magmas (Fig. 17), which can lead to disruption and rotation of tube segments. Broken and rotated tube segments range to tens of meters and thus were probably significantly longer before disruption.

Crosscutting relationships between schlieren-bounded layers, comparable to trough cutoffs in sedimentary rocks (Figs. 2–5) and mineral grading in schlieren, provide information about younging directions in both stationary and migrating tubes and thus information about the temporal evolution of tubes. In migrating tubes, rare reversals in migration direction are sometimes preserved, but in general tube centers migrate in a single direction, both in a single tube and in adjacent tubes (Fig. 4). It is not yet clear whether these migration directions follow any chamber-wide pattern(s). In stationary tubes these crosscutting relationships indicate that tube diameters decrease with time and young toward tube centers (Fig. 3).

The compositions and textures in tubes vary tremendously, from fairly simple cases of mafic schlieren grading into more intermediate to felsic compositions that are macroscopically similar to the host magma (Figs. 2A, 2B), to very complex cases where each layer displays compositions and textures different from both nearby layers and from the host composition (Fig. 2C, 2D). In the first case, which is particularly common in the Cathedral Peak unit, the lighter layers are composed of plagioclase with lesser amounts of quartz, K-feldspar, biotite, and rare hornblende. The mafic layers contain dominantly biotite, and abundant accessory minerals with small amounts of quartz, plagioclase, K-feldspar, and hornblende. More complex compositional variations are common in tubes in the Kuna Crest and Half Dome units. Schlieren compositions and chemistries in tubes are presented in a later section.

In many cases the composition and textures of the youngest part of the tubes are similar to those of the host magma (Figs. 3–5). In examples where tube layers feather into host magmas, the felsic layers between schlieren can be traced continuously into the host magmas (Figs. 3–5) (see also Burgess, 2006; Burgess and Miller, 2008). K-feldspar megacrysts occur in some layers in tubes in the porphyritic Half Dome granodiorite and Cathedral Peak granodiorite units. In rare cases more mafic dioritic magmas, enclaves, or xenoliths occur in tubes (Figs. 4E, 4F): in just a few examples the last pulse of magma broke through the slightly older outer tube rings (Fig. 3D).

Magmatic mineral foliation(s) and less commonly lineation(s) are well defined in tubes (Fig. 5). Two types of foliations are preserved in tubes: an early foliation subparallel to tube margins, best preserved in the more mafic layers (Type 1 of Žák et al., 2007), and a younger foliation that ignores layer orientations in tubes and is parallel to regional foliations in the batholith (Type 3 or 4 of Žák et al., 2007). This latter foliation is best seen in the more felsic layers in tubes (Figs. 5 and 17D). Subvertical, layer-parallel sections appropriate to examine mineral lineations in tubes are rare. In the few examples located, lineations are best displayed by the alignment of large, euhedral hornblende crystals in the mafic base of schlieren (Fig. 5). Lineations both parallel (steeply plunging) and perpendicular (shallowly plunging) to tube axes occur.


Pipes are enclosed, sometimes funnel-shaped to more commonly cylindrical-shaped bodies, geometrically similar to tubes, but distinct from tubes in that they have no repeated internal layering and thus have a single, dominant composition distinct from and more felsic than the surrounding magma (Fig. 7). Pipes are less common than tubes and troughs in the Tuolumne batholith, but have been found in all four units: they are most common in the porphyritic Half Dome and Cathedral Peak granodiorites.

Cylindrical pipe diameters vary from a few centimeters to >10 m and maintain these diameters along their axes (Figs. 7 and 8). Pipes with diameters >~10 cm have axes consistently steeply plunging in Tuolumne batholith, and thus their maximum vertical dimensions are unknown. Where good three-dimensional (3D) exposures of pipes occur, the vertical dimensions or long axes of the pipes are always greater than their subhorizontal diameters and are at least a few meters in length (Fig. 7). Pipes with diameters <10 cm have slightly more variable axis orientations but in general still plunge steeply (Figs. 7 and 8). Less common funnel-shaped pipes have decreasing diameters with depth (Fig. 7F), although they share all other characteristics with the cylindrical pipes discussed above.

Pipes in the porphyritic Half Dome and Cathedral Peak granodiorites tend to be dominated by K-feldspar megacrysts, sometimes making up >80% of pipe minerals (see also Burgess and Miller, 2008), but still with at least minor amounts of all minerals seen in the surrounding host magma (Figs. 7 and 8). K-feldspar megacrysts in the pipes show field and thin section characteristics identical to those in the host magma around the pipes (Fig. 16B) (Vernon and Paterson, 2008a, 2008b; Burgess and Miller, 2008). These larger pipes may also include small microgranitoid enclaves and host rock xenoliths (Fig. 7). K-feldspar megacryst abundances in host magmas are neither depleted nor enriched as pipes are approached (Figs. 7 and 8).

One sample has been analyzed from a pipe in the northern part of the Cathedral Peak granodiorite. This pipe has 70% SiO2; is enriched (relative to nearby Cathedral Peak and Half Dome granodiorite analyses) in K2O (~6 wt% versus <4 wt% in regular Cathedral Peak and Half Dome granodiorite analyses), Ba (~2450 ppm versus ~1000 ppm in normal Cathedral Peak granodiorite analyses) and Sr (>600 ppm); remains unchanged in Al2O3; and is depleted in Na2O, TiO2, MgO, CaO, P2O5, Rb (~150 ppm), Zr, Y, and La. It has a normal distribution of rare earth elements (light, LREE enriched compared to heavy, HREE) and no Eu anomaly.

Some pipes preserve a single, weakly developed schliere-like layer, rich in biotite, along the outer pipe margin (Figs. 7C, 7D). A close examination of these biotite-rich layers shows that they are deflected around phenocrysts in the pipes, indicating that some compaction, presumably during filter pressing, occurred during layer formation (Figs. 16A, 16B). Compaction of crystals in the pipes, and thus melt removal, is also indicated by the indentation of one phenocryst by another (e.g., Fig. 16B) leading to melt assisted dissolution of zoning (Park and Means, 1996).

Magmatic mineral fabrics in pipes tend to be weaker than those seen in the host magmas or in tubes and troughs. Type 1 margin-parallel fabrics are less common and regional type 3 or 4 fabrics more typically overprint the pipes (see fabric nomenclature in Žák et al., 2007). In rare cases subhorizontal fabrics, formed at high angles to pipe margins and to regional fabrics, suggest that late vertical compaction may have aligned crystals within the pipes. Appropriate surfaces to observe mineral lineations along pipe margins have not yet been found.

A few of the largest pipes have migrating tubes preserved along the pipe margins (Fig. 8). The presence of these migrating tubes raises the question of whether formation of the pipes sometimes caused complex flow along their margins, resulting in tubes.


Troughs are open and thus asymmetrical, schlieren-bounded channels with curvatures commonly less than tubes (Fig. 9). As trough curvatures decrease, they grade into and are commonly associated with layered schlieren zones (Žák and Paterson, 2005; Burgess and Miller, 2008). Trough widths vary from a few centimeters to hundreds of meters, but many are in the 1 m to 20 m range. Trough amplitudes are typically no more than 1 or 2 m, but can sometimes reach the 10 m scale. Where 3D exposures of troughs were found, trough long axes were always much greater than trough heights and widths (Fig. 9). Locally schlieren in troughs are offset by magmatic faults, which in turn are truncated by younger troughs (Fig. 9A).

Troughs in the Tuolumne batholith are also spatially clustered in that they commonly occur in zones of multiple troughs separated by zones where few if any are preserved. They are common along the main internal contacts and local sheeted zones in the Tuolumne batholith, but also are found well away from any obvious internal contact.

Schlieren and schlieren-bounded trough orientations in the Tuolumne batholith are more complex than tubes and more difficult to evaluate because clearly defined troughs grade into fairly planar schlieren. Orientations of 585 schlieren (which include both troughs and planar schlieren) show a wide scatter, although there is a weak maximum, indicating that a number of schlieren have northwest-southeast strikes and fairly steep dips (Fig. 6C). Unlike tubes, the orientations of trough axes are highly variable in their plunges, which range from 0° to 90°. A stereonet plot of unequivocal trough axes in one domain in the main chamber shows some scatter but with many with steep plunges (Fig. 6D). However, these trough axis orientations vary tremendously from one domain to the next, such as documented (in Paterson et al., 2008) in a domain near the eastern margin of the batholith. In another domain in the Cathedral Peak granodiorite, Burgess and Miller (2008) noted that in general the schlieren strike northwest, have variable dips, and are oblique to the major internal contacts with the Half Dome granodiorite and Johnson granite porphyry units. Troughs orientations are further complicated since they also are occasionally reintruded, broken apart, and sometimes rotated by the host magma (Fig. 17).

Spectacular truncations or cutoffs of one trough by another, geometrically identical to trough truncations seen in sedimentary rocks, are common in the Tuolumne batholith; truncation angles vary from ~50° to as low as 5°–10° (Fig. 9). These truncations provide trough younging or growth directions, measured in this study as the pole to the tangent plane of schlieren at the point where the older layer is truncated (Fig. 9), that usually preserve a single direction at any one field location. It is surprising that these trough younging directions typically trend either parallel to nearby margins or toward older units (outward) rather than inward toward younger magmatic units as defined from geochronologic studies (Bateman and Chappell, 1979; Matzel et al., 2006b. 2006c). These schlieren-bounded troughs were recognized by Bateman (1992), who described them as forming an outward-facing schlieren arch (hinted at in the weak maxima of schlieren in Fig. 6). Potential causes of these unexpected trough younging directions were discussed in Paterson et al. (2008) and Žák et al. (2009).

Trough compositions are much less variable than in tubes, and except for the schlieren typically appear similar to host compositions. Schlieren in troughs show the same features as those seen in tubes and are discussed in detail in a later section. K-feldspar megacrysts occur in troughs in porphyritic Half Dome and Cathedral Peak granodiorite units (Fig. 9) and show several interesting features. They do not follow the usual mafic to felsic zoning and reverse grain size grading seen in schlieren. They also sometimes show deflection of schlieren around them, suggesting that significant compaction of the schlieren occurred around these resistant crystals (Fig. 17). Their alignment is variable from one schliere to the next, and sometimes is different than the alignment of mafic minerals in the same layer (Fig. 10).

Magmatic mineral fabrics in troughs are similar to those in tubes in that foliations vary from approximately margin parallel in mafic layers (type 1 fabrics) to regional orientations (types 3 or 4) in the more felsic interiors. Subhedral to euhedral hornblende and biotite are sometimes imbricated relative to schlieren bases (Solgadi and Sawyer, 2007; Burgess and Miller, 2008). Only a few examples of magmatic lineations have been seen in troughs: these vary from those defined by large euhedral hornblende crystals in basal schlieren layers, which are always oriented down the central axis of the trough (Fig. 10), to a steeply plunging lineation in the more felsic parts of some troughs. As noted here, K-feldspar megacrysts, when present in troughs, sometimes show excellent alignment in schlieren but much weaker alignment in more leucocratic layers (Figs. 9 and 10). In the few cases where layer-parallel surfaces in schlieren are visible in troughs with megacrysts, these megacrysts did not form obvious lineations in the foliation plane (Fig. 10C).

Within-Chamber Diapirs

A number of small-scale diapirs (within-chamber diapirs of Weinberg et al., 2001) occur in the Tuolumne batholith. These small-scale diapirs are herein defined as moderately to non-layered, irregularly shaped batches of magma, intruding host magma of a different composition (Figs. 11 and 12). Small-scale diapirs are less common than tubes and troughs in the Tuolumne batholith, but have been found in all units except the Johnson granite porphyry. All diapirs found to date occur in compositionally layered domains and/or near internal contacts, although caution is suggested since it may be particularly difficult to recognize diapirs in non-layered regions if they are compositionally similar to their host (e.g., see plume discussion). Diapir shapes are highly irregular, but often include narrow tails and bulbous or mushroom-shaped heads where the diapirs are largely separated from their origin (mature diapirs) and are more cylindrical if still connected to layers (immature diapirs). The size of within-chamber diapirs recognized to date in the Tuolumne batholith does not exceed several meters.

Diapir compositions are highly variable relative to both the immediate host rock and to other diapirs. Diapirs in the Tuolumne batholith are sometimes slightly more mafic or more felsic than host magma, and sometimes rich in mafic minerals, K-feldspar megacrysts, and enclaves (Figs. 11 and 12). Internal layering can be present or absent but is typically less prominent than in tubes and troughs.

Several features provide information about the movement directions of these diapirs: (1) the presence of diapir tails and heads (Figs. 11 and 12); (2) the intrusion into and resulting deflection of older features (Figs. 11A, 11D); (3) the accumulation of mafic minerals (mostly biotite) to form weak schliere on margins of the diapir margin head (Fig. 11B) and not along the tail, which probably reflects filter pressing of melt and accumulation of residual minerals during diapir movement (Weinberg et al., 2001). Rarely do diapirs in the Tuolumne batholith record simple vertical movement: examples of diapirs moving up, horizontally (Figs. 16A, 16C, 16D), obliquely (Fig. 16B), and down relative to present-day horizontal (Figs. 11 and 12) are all preserved (Žák and Paterson, 2005). Diapirs moving downward relative to present-day horizontal (Fig. 12) are herein referred to as drips (Bergantz and Ni, 1999).

Plume Heads

Weinberg et al. (2001) described elliptical, schlieren-bounded, magmatic ellipsoids or thermal plume heads in the Tavares pluton, Brazil. Plume heads are particularly common in the Cathedral Peak unit and rarely occur in leucogranites in the Tuolumne batholith such as the Johnson granite (Fig. 13), although some of the characteristics of those in the batholith differ from those described in the Tavares pluton. Elliptical, schlieren-bounded regions with fairly homogeneous internal compositions, often similar to the host rock compositions, define plume heads in the Tuolumne batholith. The most common compositional difference is the reduction or lack of K-feldspar megacrysts in the ellipses, resulting in a slightly more mafic composition. Only rarely is more than one schliere ellipse preserved in a single plume head (e.g., Fig. 13). Thus these features are distinct from tubes in the lack of multiple rings, the faintness of the schlieren layers, and the similarity in compositions with the host magmas. Weinberg et al. (2001) noted that the 3D shapes of plume heads in the Tavares pluton were typically ellipsoidal. This is sometimes true in the Tuolumne batholith, but in other cases, dimensions in one direction were significantly greater than the other two, resulting in elongate bodies geometrically closer to those described as pipes or tubes in this paper.

One cluster of 51 plume heads was examined in the Cathedral Peak granodiorite near Tuolumne Meadows (Figs. 6E, 6F, and 13). These all have fairly weak schlieren rings and macroscopic characteristics (composition and texture) that closely match the host magma characteristics. They were best exposed in subhorizontal surfaces and only in a few cases was it established that their longest axes were steeply plunging approximately parallel to the steep mineral lineation seen throughout the Tuolumne batholith. It is thus assumed that in the approximately horizontal surface the ellipses reflect intermediate and short axes: intermediate axes ranging from 17 cm to 6.3 m and ellipse ratios range from 1.0 to 2.9 (average = 1.7). The intermediate tube axes show a wide range in orientations with a potentially weak correlation to two magmatic fabrics measured in this domain (Fig. 6). However, there is certainly no strong alignment of intermediate tube axes.

A number of other shapes probably related to plume heads also occur in the Tuolumne batholith. Elongate shapes with small protrusions (Fig. 13C), mushroom- or jellyfish-shaped bodies that grade into the surrounding matrix (Figs. 13D, 13E), and diapir-like heads attached to dike-like stems or tails (Fig. 13F) all occur. Weak schlieren layers bound the heads of these bodies but tend to disappear at the bottoms (Fig. 13D) or tails (Fig. 13F). Slices through the heads of all of these structures appear identical to plume heads, as would the compositions. However, the 3D shapes are more complex and merge into those described above for diapirs. They are distinct from these diapirs in that their compositions more closely match host magma compositions and that there is no evidence that they formed in compositionally layered regions.

Combinations of Structures

It is not uncommon in the Tuolumne batholith to find combinations of the above structures at a single locality and field evidence that these structures formed at the same time. Figure 14 displays some typical examples. K-feldspar megacryst–rich, funnel-shaped pipes sometimes have their tails merging into schlieren troughs (Fig. 14A), suggesting that the K-feldspar-rich magmas were separating from these mafic-dominated layers in the troughs. Sometimes these pipes feed into troughs with no K-feldspar megacrysts, even though adjacent portions of the troughs do have some megacrysts (Fig. 14A). However, this is not a general rule. K-feldspar accumulations identical to those found in some pipes are also found in troughs (Fig. 14B). K-feldspar megacryst–rich clusters, sometimes forming pipes, are sometimes associated with either migrating or stationary schlieren tubes (Figs. 8, 14C, and 14D) (see also Burgess and Miller, 2008). The close association of these K-feldspar-rich and K-feldspar-poor structures suggests a potential genetic link, a possibility I explore in a later section.

Diapirs are common in layered zones, which sometimes consist of numerous schlieren-bounded troughs (Fig. 11D). Examples in the Sawmill Canyon area (Fig. 1) were discussed in Paterson et al. (2008).

Tubes, pipes, plume heads, and diapirs all involve the buoyant movement of lower viscosity magmas through a higher viscosity host magma. When combined with the potential gradations between these structures (presented in previous sections), these observations raise the issue of whether there may be genetic links between them. For example, tubes, pipes, plume heads, and diapirs show gradations in characteristics.

Nature of Schlieren

All five of the magmatic structures discussed in this paper may be associated with schlieren either within or along their boundaries: thus the nature of the schlieren is important in understanding the development of the structures. Two distinct types of schlieren occur: those consisting of well-developed, repeated, graded layers in tubes and troughs (Figs. 2–10), and the much fainter biotite-rich layers discontinuously formed along margins of some pipes, plume heads, and diapirs (Figs. 3–14). The former, herein called graded schlieren, have received a fair amount of attention in both the Tuolumne batholith (e.g., Barbarin et al., 1989; Bateman, 1992; Reid et al., 1993; Solgadi and Sawyer, 2007; Burgess and Miller, 2008) and in many other plutons (Lykhovich, 1964; Barriere, 1981; Trent, 1981; Hanson and Nash, 1996; Antipin et al., 1997; Ventslovaite, 1998; Wiebe and Collins, 1998; Murray et al., 1993; Kagashima, 1999; Stallings and Hogan, 1999; Preston et al., 2000; Gibson et al., 2001; Weinberg et al., 2001; Clarke, 2003; Getsinger, 2004). The latter schlieren, herein called weak schlieren, have not been rigorously studied in the Tuolumne batholith, but appear identical to schlieren in the Tavares pluton discussed by Weinberg et al. (2001).

In the Tuolumne batholith, graded schlieren are centimeters to tens of centimeters in width and typically have sharp bottoms and diffuse tops. Reverse size grading of minerals is the norm, but with some exceptions, such as examples where K-feldspar megacrysts and large euhedral hornblende phenocrysts occur at the base of schlieren. As the sharp bases of schlieren are approached, a modal increase in biotite and hornblende and an unusually high increase in accessory minerals, including zircon, sphene, apatite, rare allanite, and oxides, occurs (see also Reid et al., 1993; Burgess and Miller, 2008; Paterson et al., 2008). This dramatic increase in accessory minerals is a common feature in schlieren worldwide (Lyakhovich, 1964; Hanson and Nash, 1996; Antipin et al., 1997; Ventslovaite, 1998; Murray et al., 1993; Kagashima, 1999; Stallings and Hogan, 1999; Preston et al., 2000; Gibson et al., 2001; Getsinger, 2004).

Figure 151502 compares geochemical analyses of graded schlieren in the Tuolumne batholith (black symbols) to analyses of the main units in the batholith (colored symbols). Whether these geochemical results are in part due to fractionation by crystal compaction and filter pressing during the formation of schlieren or in the development of compositional diversity in general in the Tuolumne batholith is debated (Bateman and Chappell, 1979; Reid et al., 1993; Glazner et al., 2004; Burgess and Miller, 2008). However, field evidence of compaction and removal of melt associated with schlieren and other structures is present in the Tuolumne batholith (Fig. 16). For example, the bending of schlieren around large crystals (Fig. 16A), indentation of one crystal by another (Fig. 16B), subhorizontal foliations at high angles to steep tube walls (Fig. 8B), and veins of felsic melt rising off particular domains, that form “dish and pillar-like” structures (Figs. 16C, 16D), provide clear evidence of compaction and filter pressing. Our geochemical analyses (Fig. 151502) clearly support observations noted by Reid et al. (1993), that the chemistry of the schlieren moves dramatically away from host magma compositions, do not follow the trends seen in the main units, and are compatible with mixing plus crystal-liquid fractionation of certain minerals (see discussion). Initial results for any given SiO2 value support depletion in Al2O3, Na2O3, and enrichment in Rb, MgO, CaO, K2O, TiO2, P2O5, Zr, Y, La, and Ce (Fig. 151502).

Solgadi and Sawyer (2007) noted that whole-rock geochemical analyses of graded schlieren in the Sawmill Canyon area (Fig. 1) show trends compatible with the redistribution of mafic and accessory minerals, but have different geochemical trends than seen in nearby nonlayered granodiorite. They also noted that microprobe analyses of hornblendes from these schlieren define three distinct populations that composi-tionally match those seen in the three main units of the Tuolumne batholith (Kuna Crest unit, Half Dome granodiorite, Cathedral Peak granodiorite). Furthermore, no systematic decrease in Mg number occurred from bottom to top of schlieren, which Solgadi and Sawyer (2007) concluded argued against simple fractionation of a single magma to form these layers: they concluded that mechanical erosion, mixing of crystals, and redeposition played a dominant role in schlieren formation.

Paterson et al. (2008) also examined graded schlieren in tubes and troughs and layered sequences in the Sawmill Canyon area. Major and trace element analyses for schlieren, normalized by the typical composition of the nearby Half Dome granodiorite (the likely source), show that with decreasing SiO2 a more than threefold enrichment in Ti, Fe, and P occurs, with less pronounced increases in Mn and Mg (Fig. 151502). The contents of Al, Na, Ca, and K are largely unchanged. Among trace elements, the most pronounced changes are enrichments in Zr and Hf (>4× ), followed by LREE and middle (M) REE (~3.5×), Nb and Ta (close to 3×), Th, U, and HREE (slightly above 2×). The contents of K, Sr, Na, K, Al, and Rb remain almost constant. We (Paterson et al., 2008) concluded that such a variation is compatible with progressive accumulation of mainly mafic and accessory phases, most notably amphibole (Fe, Mg, Ti), titanite (Ti, Nb, Ta, U, Th, LREE), magnetite (Fe, Ti, Mn), apatite (P, MREE), and zircon (Zr, Hf, and HREE) with a lesser role for feldspar(s), a conclusion consistent with what is seen in the field and in thin sections of schlieren. Our data confirmed the notion that the origin of schlieren cannot be due to a simple accumulation of crystals during fractional crystallization. Instead, magma mixing and/or mingling and/or erosion and recycling of material from older pulses must be invoked (Reid et al., 1993). This conclusion is supported by least squares modeling in which all calculations required little or no preferential accumulation of feldspars, whereas the accumulation of amphibole + biotite + magnetite > titanite > apatite + zircon is important (Paterson et al., 2008).

Burgess and Miller (2008) examined graded schlieren within the Cathedral Peak granodiorite and concluded that they represent locations in the batholith where crystals have been segregated in some fashion from melt, and thus are cumulates. They noted that the schlieren in some cases have extraordinarily high REE contents and much higher Zr contents (from accumulated zircon) than average Cathedral Peak granodiorite (e.g., Reid et al., 1993; Miller et al., 2007). For example, analyses from two schlieren are characterized by overall REE enrichment reaching 800–1000× chondrite for LREEs in the most mafic schlieren. One sample contained >1500 ppm Zr, >60 ppm Nb and Y, >400 ppm Ce, and >100 ppm Th, and is correspondingly depleted in feldspar compatible elements and quartz. Schlieren they analyzed also display very slight negative Eu anomalies. Burgess and Miller (2008) noted that anorthite content in mafic schlieren increased to An48 (versus the highest An content of An39 in nearby units) and had more variable core compositions that those from other Cathedral Peak samples.


Timing of Formation of Structures

I noted that because all 4 major units preserve examples of tubes, troughs, pipes, and diapers, and because these units have well-established crystallization ages ranging from 95 to 85 Ma, the formation of these structures must be time transgressive over ~10 m.y. This is because it is highly unlikely that any part of the batholith formed and stayed above its solidus for longer than a few million years, a conclusion supported by thermal modeling of and cooling ages from the Tuolumne batholith (Mundil et al., 2004; Matzel et al., 2006b, 2006c; Paterson et al., 2007b). Direct evidence of the time- transgressive nature of these structures is found along the main internal contacts, where tubes and troughs in older units are intruded and truncated by the younger intrusive phases (Fig. 17).

Additional observations establish the development at any one location of these structures relative to the magma solidus and other structures. (1) All crystals in these structures are dominated by magmatic characteristics with only very localized and minor microstructural evidence of subsolidus alteration (for summaries see Moore and Sisson, 20008; Vernon and Paterson, 2008a, 2008b). (2) The concentration of magmatic minerals in schlieren requires that a large percentage of the crystals were already present during flow and crystal sorting. (3) Some layers, particularly in tubes and troughs, grade into the host magmas (Figs. 3–5 and 9). (4) A number of the structures were reintruded by host magmas, during which time the structures were sometimes broken into segments and/or rotated (Fig. 17). (5) Most if not all of these structures are partially or completely overprinted by late magmatic fabrics (Fig. 17D) (Žák et al., 2007; Paterson et al., 2008).

These observations indicate that the above structures formed by flow of crystal-bearing magmas through hosts of crystal-rich magma mushes. The structures also formed before formation of final magmatic fabrics and thus above, but probably fairly near, the solidus of the magmas in the structures. Field observations also indicate that the tubes, troughs, and pipes formed at the same time at any one locality. This conclusion implies that a great deal of localized flow, resulting in differentiation and formation of compositional and structural complexity, can occur in crystal-rich mushes (>50% crystals) but still above the solidus.

Rheological Characteristics of Magmas during Formation

Weinberg et al. (2001) argued that marginal schlieren and extraction structures, and consequently ladder dikes, snail structures, diapirs, and plume heads, are ideally developed when the magma consists of a permeable crystal mush. They specifically noted that the presence of similar structures in K-feldspar megacryst-bearing granites of the Sierra Nevada suggests that these structures may best be developed and preserved in high-K, calc-alkaline granite-granodiorite with heterogeneously distributed K-feldspar megacrysts (in general 15%–25% modal megacrysts). One implication of my study, besides establishing that similar structures formed in magmas without K-feldspar megacrysts, is that they formed in higher crystallinity magmas than suggested by Weinberg et al. (2001), that is, in magmas with fairly high effective viscosities (Bergantz and Ni, 1999, Bergantz, 2000; Walker et al., 2007). The following observations in the Tuolumne batholith address specifically the above hypothesis. (1) When structures are reintruded, broken apart, and/or rotated by host magmas, the structures act as rigid objects and do not completely collapse (mingle) or deform internally (Fig. 17). (2) When layers in structures were cut by magmatic faults, the movement on these faults rigidly displaced but did not internally disrupt these layers (e.g., Fig. 9). The same is true when structures at internal margins were intruded and truncated by younger pulses, which required partial removal of the structure without internally deforming the remainder (Fig. 17). (3) Many troughs formed during magmatic erosion of older layers prior to renewed deposition to form new layers: this erosion removed parts of the older layer without otherwise deforming the remaining layering (Fig. 9). (4) All of these structures form during intrusion of one magma pulse into another along relatively sharp contacts across which no obvious mingling and only rare mixing has occurred (Figs. 3–17). The formation and preservation of these sharp contacts suggests that at least the host rock magmas must have been fairly strong due to high crystal percents (Saar et al., 2001; Bergantz, 2000). (5) Compaction of magmas and indentation of crystals (Fig. 16) during transmission of deviatoric stresses through magmas also requires that the magmas are fairly crystal rich.

All of these observations support the interpretation that the magmatic host rocks were fairly strong and thus crystal rich, at least at the time scales and rates at which the structures formed, although these host rocks must have retained melt since they (1) continued to contribute magma to the developing structures, (2) flowed magmatically during reintrusion of some structures, and (3) strained in a magmatic state during the hypersolidus formation of magmatic fabrics (Žák et al., 2007). Magmas in the developing structures presumably began with fewer crystals than the host magmas, but still had a significant percentage of crystals, since all crystal phases were sorted during formation of the structures and they had to have enough strength to preserve steeply dipping layering, resist subsequent erosion, undergo compaction and filter pressing, and act as rigid objects during reintrusion and rotation.

Processes by Which Structures Formed

The formation of all of these structures involved multiphase magmatic flow during initially high and then decreasing temperatures and increasing crystal percents. Unfortunately, multiphase flow, particularly during changing conditions resulting in crystal growth, is not well understood (e.g., Carrigan et al., 1992; Bergantz, 2000; Bergantz, and Breidenthal, 2001; Petford, 2003; Burgisser et al., 2005; Dingwell, 2006). In spite of this limitation, the existing field observations and geochemistry place fairly large constraints on the development of these structures.

Weinberg et al. (2001) suggested that in the Tavares pluton tubes represent cylindrical channels of magma flow through crystal mushes, and that the main flow direction is parallel to the tube walls (Fig. 18). They interpreted slightly (snail structures) to fully migrating tubes (ladder dikes) as similar structures that result from the superposition of sequential cylindrical magma pathways in which each new, curved schliere represents the walls of a former cylindrical magma path. Magma flow through tubes, potentially either up or down, is driven by thermal or compositional buoyancy (e.g., Griffiths, 1986; Martin et al., 1987; Weinberg et al., 2001; Bachmann and Bergantz, 2008). Schlieren along the tube walls form by combined flow sorting and filter pressing.

This model explains well the tubes in the Tuolumne batholith with the following refinements. In the Tuolumne batholith, the compositions and textures of magma that flowed through the tubes can be quite variable, implying that different magma sources were tapped. Also both stationary and most migrating tubes decrease in diameter with time, an observation interpreted to reflect decreasing flow velocities and thus shrinking tube diameters. Mineral fabrics in tubes supports vertical motion parallel to tube margins, sometimes followed by vertical compaction in tubes (as flow ceased?), and finally regional strain resulting in the development of overprinting magmatic fabrics. The locally consistent direction of tube migrations raises the question of whether source regions of rising magma are migrating in consistent directions or if lateral flow of host magma past a fairly steady magma source may displace preexisting tubes (a hotspot analogy). Observations in the Tuolumne batholith indicate that the host magma to tubes was stiff enough over the time scales of tube formation to resist mingling and collapse of the tube, but weak enough that it could move into and through tubes, intrude through tube walls, or reintrude and break apart tubes.

Troughs in the Tuolumne batholith are considered to be analogous to sedimentary flow channels in which the channels have a bedload and formed in a porous media (Fig. 18) (e.g., Wahraftig, 1979; Solgadi and Sawyer, 2007; Dufek and Bergantz, 2007). These channels in the Tuolumne batholith vary from fairly localized with similar width/depth ratios to broad gently curved surfaces (large width/depth ratios) probably associated with broad sheet flows along an internal crystal mush boundary. Magma flow is probably parallel to trough axes, as inferred by aligned hornblendes in basal schlieren. Host magmas, including earlier troughs, were stiff enough to allow local erosion followed by redeposition and associated filter pressing (see Nature of Schlieren discussion) and local lag deposits to form (Fig. 14). The similar material in the troughs and in the host magmas suggests local sources: the highly variable orientation of trough axes opens the door to variable processes driving magma flow in troughs, such as crystal mush avalanches, gravity-driven channelized flows, convection driven flow along irregular mush zone margins, and downward return flow during ascent of new magmas. The most intriguing aspect of Tuolumne batholith troughs is the common outward younging of channels. One possible explanation is that stresses caused by local processes (Paterson et al., 2008), regional tectonic processes (Paterson et al., 1998), and by the cooling of magma chambers (Žák et al., 2009) combine to produce outward-migrating low stress sites favorable for local redistribution of late melts in chambers.

Two mechanisms for the formation of pipes have been proposed in the literature. Wiebe (1996; also see Wiebe and Collins, 1998) suggested that immediately after the juxtaposition of magmas with different compositions, the less dense magma rises up through the more dense magma, largely as a Raleigh-Taylor instability. This would result in pipes with variable compositions and grain sizes relative to the host magmas, comparable to some of the pipes seen in the Tuolumne batholith. Alternatively, Weinberg et al. (2001) noted that coarse K-feldspar aggregates occur on the upstream side of dike necks, as inferred from crosscutting relationships in the dike. They suggested that flow necking led to a megacryst “logjam,” which continued to grow by the filtering out of megacrysts upstream (see also Clarke and Clarke, 1998). If these “logjams” occurred in local channels such as tubes, troughs, and diapers, they would result in K-feldspar–dominated pipe-shaped regions (Figs. 7, 8, and 18), as seen in the Tuolumne batholith. Caution for this interpretation is recommended because K-feldspar accumulations in the Tuolumne batholith have highly variable geometries, many not being associated with tubes and troughs, implying that there may be multiple ways to form them (Vernon and Paterson, 2008a). However, one common theme in all these accumulations is evidence of compaction of the megacrysts and removal of interstitial melt (Fig. 16), suggesting that filter pressing is an important process during their formation.

The above processes all involved faster channelized flow of magma through host magma that is stationary or moving slowly relative to the channel flow. Thus the channel margins may be eroded but not otherwise displaced by the channelized flow. In contrast, diapirs and plume heads are interpreted to be magma batches that moved through host magma that was displaced around the diapir and potentially incorporated into plume heads. The thin tails, broad heads, and compositions distinct from host magmas of most Tuolumne batholith diapirs indicate that they were detached (typically from unobserved sources) batches of magma moving through but not significantly incorporating host magma (Marsh, 1982; Clemens, 1998; Miller and Paterson, 1999; Olsen and Weeraratne, 2008). They are nicely explained as Rayleigh-Taylor instabilities (Berner et al., 1972; Marsh, 1982; Ronnlund, 1987; Whitehead and Helfrich, 1991; Weinberg and Podladchikov, 1994; Dietl and Koyi, 2002; Olsen and Weeraratne, 2008) and since they move through hot, crystal mushes can potentially ascend long distances relative to their sizes (e.g., Marsh, 1982; Weinberg and Podladchikov, 1994). A few have less well defined heads and broader dike-like stems still attached to source layers, and thus fit the well-known characteristics of immature Rayleigh-Taylor diapirs (Berner et al., 1972; Dixon, 1975; Whitehead and Helfrich, 1991; Olsen and Weeraratne, 2008).

Griffiths (1986) showed that thermal plume heads grow during ascent by entraining their surroundings through buoyancy (heat) diffusion away from the plume head, which heats up a surrounding boundary later (Weinberg et al., 2001). The entire hot boundary layer is potentially entrained, in which case the plume-head buoyancy may remain constant even though its volume increases and the plume cools. This contrasts with compositionally driven diapirs, which do not easily entrain surrounding magma because of the slower chemical diffusion as compared to thermal diffusion (Weinberg et al., 2001). These observations explain well the characteristics of Tuolumne batholith magma ellipsoids with compositions similar to those of host magmas and bounded by weak schlieren. However, their variable shapes, particularly those grading into sheet-like bodies (Fig. 13), suggests that these magmas may first move by a more channelized flow process and then form plume heads as their flow rates decrease.

For example, Olsen and Weeraratne (2008) presented an intriguing set of analogue experiments of metal-silicate plumes and/or diapirs and “emulsion diapers,” an interesting analogue for crystal-mush systems. Even though their study focused on a different environment (sinking of metal through mantle to form the Earth's core), and different materials (liquid gallium sinking through corn syrup with no thermal effects), their experiments produced a number of features remarkably similar to those seen in the Tuolumne batholith. Olsen and Weeraratne's (2008) experiments produced (1) diapir and plume heads, often attached to pipe-like paths that had significant life spans in their models and sometimes continued to focus channelized flow of material (cf. their Figs. 1 and 2 to the tubes and pipes in the Tuolumne batholith); (2) half diapirs (cf. their Fig. 6 and Fig. 14 herein); (3) a means of mixing of host material into the diapirs and pipe-like structures; and (4) cases where layer instabilities led to the above structures (their Figs. 4 and 5). Furthermore, they presented examples where the channelized movement of plumes and emulsion diapirs (their Figs. 7 and 8) resulted in thermochemical plumes forming along the borders of the features produced by their experiments, plumes similar to those seen along the borders of pipes in the Tuolumne batholith.

The most unusual aspect of diapirs and plume heads in the Tuolumne batholith is their highly varied movement directions, indicating that gravity and/or buoyancy is not the dominant driving mechanism. Buoyancy must have operated, but was dominated by other poorly constrained gradients in the host magma such as gradients in effective viscosities, differential stresses, or host magma flow (Žák and Paterson, 2005).

Many of the above structures are associated with bounding and/or internal schlieren, which in this paper were grouped into “graded” and “weak” schlieren. A number of processes have been proposed for the origin of schlieren in granitoids: (1) partial assimilation of mafic enclaves, (2) crystal settling, (3) shearing out of inhomogeneities, (4) steep physicochemical gradients at flow margins (e.g., Barriere, 1981) leading to preferential crystallization of ferromagnesian minerals and suppression of crystallization of felsic minerals (Naney et al., 1980), and (5) shear sorting against an effectively rigid wall (Bhattacharji and Smith, 1964; Komar, 1972; Barriere, 1981). In the Tuolumne batholith, schlieren in these structures all share the following: (1) they formed along boundaries of differential flow; (2) these boundaries varied from vertical to horizontal during schlieren formation; (3) size grading (of selective mineral populations) occurs with distance from schlieren bases in some schlieren, but not all; (4) dramatic compositional grading occurs with distance from schlieren base and particularly formed by the accumulation of mafic (biotite and hornblende) and accessory (sphene, zircon, apatite) minerals into the basal zones; (5) mineral alignment that is subparallel to slightly oblique at bases to highly oblique at schlieren tops; (6) evidence for mineral compaction and removal by filter pressing; (7) formation in crystal-mush systems, in which all mineral phases had begun to crystallize; and (8) involved already premixed crystal populations, some of which came from other parts of the magma chamber (e.g., zircon antecrysts of Matzel et al., 2007; hornblendes of Solgadi and Sawyer, 2007). Thus the proposed processes for schlieren formation must involve both physical sorting and/or alignment during flow and compositional sorting during crystallization and crystal-liquid fractionation and thus are only compatible with a combination of steep physicochemical gradients at flow margins and shear sorting against an effectively rigid wall, mentioned above.

Weinberg et al. (2001) postulated that schlieren with size sorting, such as the graded schlieren, may result from the combined effects of shear flow and loss of interstitial melt (by filter pressing) to the permeable magmatic walls (Fig. 18). They noted that this requires a negative pressure gradient toward the walls, resulting from unbalanced pressures on a wider scale (beyond the local flow observed), but could not rule out other gradients driving melt migration into the structures. They also argued that shear sorting and melt loss are most effective when the walls enclosing the flow are a crystal-liquid mush behaving as a permeable, viscoplastic material that not only favors the formation of schlieren, but favors melt extraction from mush pores. This model is generally compatible with Tuolumne batholith graded schlieren formation, which our initial least squares geochemical modeling suggests is compatible with progressive accumulation of amphibole + biotite + magnetite > titanite > apatite + zircon with little role for feldspar(s) (Paterson et al., 2008; Žák et al., 2009). These data also indicate that magma mixing and/or mingling and/or erosion and recycling of material from older pulses must be invoked either before or during schlieren formation (see also Reid et al., 1993).

The weak schlieren in the Tuolumne batholith are not yet well studied. Their common setting at the outer boundaries of diapirs, plume heads, and pipes, and their greatest intensity along boundaries in the direction of inferred movement of these structures (Fig. 13), supports their formation by a filter-pressing mechanism in which mafic minerals are preferentially collected in the host magmas and melts driven off (e.g., Weinberg et al., 2001).


In the Introduction, I raised three related issues. (1) Could local convection occur in crystal-rich magmas? (2) How and when did mixing of crystal populations occur? (3) To what degree do local magmatic structures provide information about chamber construction versus processes that occurred within already constructed chambers?

The structures described herein clearly indicate that movement of crystal-bearing magmas may continue well below magma liquidi and fairly close to magma solidi at least in the wet granodiorites examined. Magma movement continued even after these structures were sufficiently crystallized to act as rigid objects, as indicated by their breaking apart and rigid rotations.

The details of how mixed crystal populations occurred in the Tuolumne batholith remain uncertain. However, the sorting and accumulation of crystals during formation of the structures involved magmas that already had diverse mineral populations, based on geochronologic (Matzel et al., 2005, 2006b, 2007; Miller et al., 2007), geochemical (Reid et al., 1993; Burgess and Miller, 2008), and field and microstructural studies (Solgadi and Sawyer, 2007; Paterson et al., 2008); thus a great deal of crystal mixing occurred before and/or during the formation of these structures. Since this mixing involved crystals from all main magmatic units in the Tuolumne batholith, mixing must have occurred as localized flows passed through other units either in a vertically or laterally extensive chamber. Such mixing is difficult to imagine in a system formed of small pulses that did not dramatically interact with one another, such as suggested by Coleman et al. (2004) and Glazner et al. (2004). The timing of mixing also has implications for volcanic processes. The complex crystal cargoes seen in erupted units already existed in the Tuolumne batholith and need not have formed during eruption.

Ruprecht et al. (2008) presented an exciting study of crystal mixing processes, particularly mixing during magma overturn driven by gradients in gas bubbles, but also applicable to thermal or compositionally driven overturn: they experimentally produced structures during mixing and overturn (their Fig. 1) that look identical to the diapirs, plume heads, pipes, and tubes described in the Tuolumne batholith. Ruprecht et al. (2008) concluded that even during a single overturn, crystals that were originally hundreds of meters apart can be juxtaposed at the centimeter scale, and after multiple overturns, crystals from diverse parts of a chamber can be assembled. They further concluded that in spite of magma overturn, some stratification in chambers is typically preserved, which may explain the variable magma types seen in the Tuolumne batholith tubes.

Another implication of the existence of diverse crystal populations in the Tuolumne batholith concerns the inferred residence times of crystals in magmas. Studies of diverse crystal populations in volcanic deposits use crystallization and/or resorption rates during disequilibrium to establish potential residence times of crystals in subvolcanic chambers. However, in the Tuolumne batholith there is good evidence that complex mineral populations existed that were not in chemical or textural equilibrium and had mineral ages varying by millions of years (e.g., Matzel et al., 2005, 2006b, 2006c; Miller et al., 2007). A likely solution is that some crystals were armored (rimmed) by others or existed in crystal clusters that prevented rapid chemical communication with new melts or crystals with which they were in disequilibrium. Thus caution is urged regarding existing conclusions about the length of crystal residence times in subvolcanic chambers.

This study (and Žák and Paterson, 2005; Paterson et al., 2008) indicates that a complex array of compositionally and textural defined structures formed in an already amalgamated magma chamber. Given the large number and simultaneous formation of these structures it is intriguing to speculate that together they form a network of increased permeability in a crystal-rich, lower permeability, host magma, and thus aided in the redistribution of magma and in crystal mixing. Whether the Tuolumne batholith chamber “boiled” in the sense of forming volatile-rich channels in the chamber (Ruprecht et al., 2008), whether blocks of crystal mushes or crystal avalanches moved through the chamber, or whether crystal-bearing currents flowed along crystal-mush margins need further study. However, all of these processes imply operation in a large, mushy, evolving chamber.

I also indicate here how important it is to recognize and look beyond the structures described herein when attempting to locate features that might give us clues about how the chambers originally formed. This is because structures such as tubes, troughs, and even local sheeted zones (Paterson et al., 2008; Žák et al., 2009) can form in an existing magma chamber and thus not record the chamber construction process. The same has been argued for preserved magmatic fabrics that in many cases may provide information about regional strain during chamber evolution, or strain caused by local flow gradients in the chamber, but not about initial chamber growth (Paterson et al., 1998; Žák et al., 2007).

A final important implication of these structures is that they challenge aspects of the incremental chamber growth model championed by Coleman et al. (2004, 2008), and Glazner et al. (2002, 2008a, 2008b) for the construction of the Tuolumne batholith. These authors have suggested: (1) that the batholith “began its life as a large [vertical] dike swarm“ (Glazner et al., 2002, p. 269), an interpretation they have recently changed to the interpretation that it “was assembled as downward-stacking laccoliths” (Coleman et al., 2008, p. 23); (2) that “Early increments cool below the solidus quickly” (Coleman et al., 2008, p. 23); “magma is 50% crystallized (and thus no longer mobile)” (Glazner et al., 2004, p. 6), “in situ crystal fractionation and/or magma mixing cannot account for the zonation” (Glazner et al., 2004, p. 7), and that “the geochemical variation of the suite reflects regular changes in the composition of magmas generated at the source” (Coleman et al., 2004, p. 435–436), all of which imply that the compositional diversity in the Tuolumne batholith could not form at the emplacement site; and (3) that subsolidus internal annealing and/or recrystallization led to removal of contacts and thus “the general rarity of chilled margins in granites, and the cryptic character of their internal contacts, to late-stage textural modification that obscures much of the record of pluton assembly” (Glazner et al., 2008a, p. 10337).

The presence of chamber-wide, knife sharp, and gradational internal contacts in the Tuolumne batholith (Žák and Paterson, 2005; Žák et al., 2007), plus the thousands of internal magmatic contacts associated with the structures, indicates that internal contacts are widespread in the Tuolumne batholith, are not cryptic nor annealed, and often occur between magma pulses with clearly defined textures, bulk compositions, and chemistries (see also Moore and Sisson, 2007, 2008; Vernon and Paterson, 2008a, 2008b). Most of these internal contacts are formed during local flow in an existing chamber, not during chamber construction or during thermal maturation and annealing. Many of these local contacts, such as defined by schlieren, have widely varying orientations (Fig. 6) that are not compatible with either a vertical sheeted dike model or a subhorizontal laccolith model, unless these models led to large magma chambers in which internal contacts were destroyed by continued movement of magma prior to formation of the structures discussed in this study.

All the structures described in this paper also provide wonderful examples that continued movement of magma at the emplacement site led to both crystal mixing and formation of compositional and structural diversity. The magnitude of these processes is difficult to evaluate at the scale of the entire batholith, but they did occur. These observations do not exclude the possibility of incremental growth of the Tuolumne batholith, although the location (deep in the crust, during ascent, during chamber growth) of the amalgamation of pulses and the number and size of pulses remain poorly constrained. However, these observations imply that by whatever means, the batholith grew fairly large magma chamber(s) in which magma pulses moved through existing crystal mushes and dispersement and/or mixing of crystals occurred.

Use of Structures as Tools for Understanding Magma Chamber Evolution

These structures provide a great deal of information about the local evolution of magma chambers. For example, methods of determining paleovertical in chambers have long been sought in paleomagnetic studies and have been used to support chamber growth models (e.g., Wiebe and Collins, 1998). In the Tuolumne batholith (Frei, 1986), the tubes and pipes provide reliable indicators of paleovertical when not broken apart during reintrusion, but troughs, diapers, and plume heads do not. Statistical averages of the plunge of tubes and pipes in the Tuolumne batholith are consistently 90° ± 2°, suggesting that the batholith was not tectonically tilted and internal parts of the chamber were not magmatically rotated during growth (Frei, 1986).

The consistent direction of tube migration at single outcrops is also intriguing and may provide information either about the direction of host magma flow in the chamber (if a hotspot analogy is correct) or about some as yet unknown process that caused tube source regions to unidirectionally migrate. I am not aware of any study that has yet examined tube migration patterns at the batholith scale. Whether such patterns are random or ordered will be of great interest in evaluating chamber evolution models.

Troughs, mineral grading, and intrusive relationships give local growth and/or younging directions, but at three different scales: mineral grading provides information at the scale of a single pulse of magma, trough cutoffs at the scale of one or more pulse(es) flowing past another, and intrusive relationship at scales ranging to large intrusive pulses. These younging directions need not match the overall younging of the chamber, but instead provide valuable information about local processes in an evolving chamber (e.g., Žák and Paterson, 2005; Paterson et al., 2008). In Paterson et al. (2008) we gave one example where an evaluation of trough younging was useful in determining that crystal mush domains in the Tuolumne batholith were being disrupted and recycled in the chamber.

The movement directions of both diapirs and plumes are surprisingly variable in the Tuolumne batholith and rarely vertical. Thus these features must be reflecting flow or rheological gradients and not just gravity, and thus provide information about these gradients.

The presence or absence of these structures in plutons may provide information about the nature of former magma chambers. For example, Weinberg et al.'s (2001) suggestion that these structures may best be developed when the magma composition is appropriate to form a porous crystal mush is an idea supported by the presence of these structures in Sierran plutons compositionally similar to the Tavares pluton. Other hypotheses worth considering are whether some magma systems crystallized so quickly that there was not sufficient time to form the structures (do they require longer-lived chambers), whether widespread magma movement may occur after formation of these structures and thus destroy them, whether channelized flow and development of permeable networks are favored at certain crustal levels, and whether subtle compositional differences make it difficult to detect these structures in other plutons. Certainly the plume heads are often easy to miss in the Tuolumne batholith, and large troughs may appear as planar schlieren zones and thus be downplayed.


1. A complex array of widespread, but domainally developed, structures is preserved in the Tuolumne batholith and includes stationary and migrating tubes, troughs, pipes, diapers, and plume heads. These structures all formed by local magma flow through crystal-mush host magmas, and are associated with the formation of mafic and accessory mineral–rich schlieren, evidence of filter pressing, and the accumulations of crystals with diverse magma histories. Together they may form a permeable network in which channelized flow occurred in the magma chamber.

2. These structures provide examples of local convection in crystal-rich magmas resulting in local compositional and structural diversity. Thus the notion that magmas must have ≤50% crystals to convect and/or fractionate at the chamber site and the ability of magmas to form compositional diversity at these crustal levels need to be revised.

3. These structures formed in an existing magma chamber and involved crystals derived from other parts of the chamber, indicating that it is necessary to find older features of the chamber to unravel chamber construction models. This is also true for preserved, chamber-wide magmatic fabrics, since these overprint the structures.

4. Interpretations derived from volcanic studies about magma residence times of crystals and crystal mixing during eruptions need to treated with caution, since there is good evidence that mixed crystal populations existed prior to eruptions, and in the case of the Tuolumne batholith involved crystals with ages ranging over 10 m.y. A likely solution is that crystals in subvolcanic chambers become armored (rimmed) by other crystals or exist in crystal clusters that, in spite of changing environmental conditions, prevent rapid chemical communication with the melts with which they are in disequilibrium.

5. The presence of these structures and resulting implications challenge aspects of the incremental chamber growth model championed by Glazner et al. (2002, 2003, 2004) and Coleman et al. (2004, 2005) for the construction of the Tuolumne batholith. I suggest that the presence and widespread nature of these structures and evidence of widespread recycling of older into younger phases are most easily explained by the existence of either laterally or vertically extensive magma chambers in which initially dispersed minerals are brought together, rather than a dike or sill construction model. Furthermore, the thousands of preserved internal contacts, some of which are easily reset, such as the weak schlieren layers dominated by biotite, provide clear evidence against late annealing and removal of internal structures. However, these structures are permissive of early pulsing leading to large chambers, followed by formation of the structures. Some, such as the compositionally diverse tubes, hint at variable sources at deeper levels in the chamber.

6. These structures are useful structural tools for evaluating the internal evolution of magma chambers. An evaluation of these structures in the Tuolumne batholith established a number of interesting results that future models must incorporate. For example, the consistently steep tube and pipe axes indicate that neither the pluton nor features in the pluton were tilted, thus excluding a model for the Tuolumne batholith in which subhorizontal layers tilted to form preserved steep contacts. The tube and pipe orientations and widespread orientations of schlieren in troughs and diapir movement directions are particularly challenging to explain in the dike and laccolith models proposed by others. Local younging directions determined from troughs indicate that outward growth occurred in a number of zones, although the overall direction of younging is toward the batholith center. The variable movement directions of diapirs and plumes require interactions between buoyancy forces and other gradients. Since all of these structures formed in a time-transgressive fashion over the duration of batholith growth, but at essentially the same time at any one locality, it will be an exciting challenge to integrate these processes in future models.

I thank Calvin Miller and Jim Moore for their very constructive reviews that significantly improved the manuscript, Robert Miller and Jonathan Miller for discussions about the Tuolumne batholith, George Bergantz and Roberto Weinberg for discussions about magmatic structures, and Stephen Holloway and Randy Keller for editorial assistance and wonderful patience. I also thank the Yosemite National Park Rangers for their constant support and interest in our work. I gratefully acknowledge support from National Science Foundation grants EAR-0537892 and EAR-0073943.